McKinsey Center for Government

Generative
AI and the
future of work
in America
July 2023

Authors
Kweilin Ellingrud
Saurabh Sanghvi
Gurneet Singh Dandona
Anu Madgavkar
Michael Chui
Olivia White
Paige Hasebe

Editor
Lisa Renaud
Cover illustration by Matt Murphy
About the McKinsey Global Institute
The McKinsey Global Institute was established in 1990. Our mission is to provide a fact base to
aid decision making on the economic and business issues most critical to the world’s companies
and policy leaders. We benefit from the full range of McKinsey’s regional, sectoral, and functional
knowledge, skills, and expertise, but editorial direction and decisions are solely the responsibility
of MGI directors and partners.
Our research is grouped into five major themes:
— Productivity and prosperity: Creating and harnessing the world’s assets most productively
— Resources of the world: Building, powering, and feeding the world sustainably
— Human potential: Maximizing and achieving the potential of human talent
— Global connections: Exploring how flows of goods, people, and ideas shape economies
— Technologies and markets of the future: Discussing the next big arenas of value
and competition
We aim for independent and fact-based research. None of our work is commissioned or paid for
by any business, government, or other institution; we share our results publicly free of charge;
and we are entirely funded by the partners of McKinsey. While we engage multiple distinguished
external advisers to contribute to our work, the analyses presented in our publications are MGI’s
alone, and any errors are our own.
You can find out more about MGI and our research at www.mckinsey.com/mgi.

MGI Directors MGI Partners

Sven Smit (chair) Marco Piccitto Michael Chui Jan Mischke
Chris Bradley Olivia White Mekala Krishnan Jeongmin Seong
Kweilin Ellingrud Jonathan Woetzel Anu Madgavkar Tilman Tacke

About the McKinsey Center

for Government
With its independent and analytical approach, the McKinsey Center for Government (MCG) is
a dedicated center of excellence that helps government leaders deliver better outcomes and
experiences for their people.
Backed by a network of global experts, MCG works alongside many of the world’s leading public
sector stakeholders and organizations to enable them to operate at the highest level.
©Eloi Omella/Getty
Contents

At a glance iv

Executive summary 1

Introduction 13

1. A robust recovery marked by job switching

and labor shortages 15

2. Job gains and losses through 2030 23

3. New forces changing labor demand:

Generative AI and federal investment 31

4. Who’s vulnerable? 43

5. Preparing for the future of work 53

Methodology brief 63

Acknowledgments 67
At a glance

— During the pandemic (2019–22), the US labor market saw 8.6 million occupational
shifts, 50 percent more than in the previous three-year period. Most involved people
leaving food services, in-person sales, and office support for different occupations.
— By 2030, activities that account for up to 30 percent of hours currently worked across
the US economy could be automated—a trend accelerated by generative AI. However, we
see generative AI enhancing the way STEM, creative, and business and legal professionals
work rather than eliminating a significant number of jobs outright. Automation’s biggest
effects are likely to hit other job categories. Office support, customer service, and food
service employment could continue to decline.
— Federal investment to address climate and infrastructure, as well as structural shifts,
will also alter labor demand. The net-zero transition will shift employment away from
oil, gas, and automotive manufacturing and into green industries for a modest net gain in
employment. Infrastructure projects will increase demand in construction, which is already
short almost 400,000 workers today. We also see increased demand for healthcare workers
as the population ages, plus gains in transportation services due to e-commerce.
— An additional 12 million occupational transitions may be needed by 2030. As people
leave shrinking occupations, the economy could reweight toward higher-wage jobs. Workers
in lower-wage jobs are up to 14 times more likely to need to change occupations than those in
highest-wage positions, and most will need additional skills to do so successfully. Women are
1.5 times more likely to need to move into new occupations than men.
— The United States will need workforce development on a far larger scale as well as
more expansive hiring approaches from employers. Employers will need to hire for skills
and competencies rather than credentials, recruit from overlooked populations (such as
rural workers and people with disabilities), and deliver training that keeps pace with their
evolving needs.

McKinsey Global Institute | Generative AI and the future of work in America iv

Web 2023
future-of-work
Exhibit 1 of 21

We expect an additional 12 million occupational transitions through 2030.

US job growth, index (0=2016 levels)

40

Resilient and growing

occupations1 30

+17%
9.9M jobs 20

Stalled but rising

occupations² 10
+7%
2.8M jobs
0

–10%
Hit and declining –6.0M jobs
–10
occupations³

2016 2019 2022 2030

Growth • Healthcare demand increase • Investments in • Automation adoption

trajectory as the population ages infrastructure and the • Sustained e-commerce trend
driven by • The push toward digitization net-zero transition • Reduced need for
and technology • Demand for reskilling and customer-facing roles
• Demand for last-mile lifelong learning
delivery

Projected 1M 1M 10M
transitions⁴ From a
to new resilient and
occupations,⁵ growing
2022–30 occupation
to any other
occupation

Occupational 36% of US workers in 2022: 25% of workers: 39% of workers:

categories • Health professionals • Builders • Production work
within each
profile • Health aides, technicians, • Creatives and arts management • Food services
and wellness • Property maintenance • Customer service and sales
Occupations • STEM professionals • Mechanical installation and repair • Office support
where • Managers
generative AI • Community services
could accelerate • Transportation services • Education and workforce
automation • Business and legal training
significantly professionals • Agriculture

1Resilient during the pandemic, 2019–22, and expected to grow between 2022 and 2030.
2Stalled during the pandemic, 2019–22, and expected to rise between 2022 and 2030.
3Hit during the pandemic, 2019–22, and continuing to decline between 2022 and 2030.
4Job transitions are defined as jobs in net declining occupations across sectors compared with the 2030 baseline.
5Even in categories that are growing overall, employment may decrease in specific occupations, requiring some workers to find new roles.
Source: O*NET; US Bureau of Labor Statistics; Current Population Survey, US Census Bureau; McKinsey Global Institute analysis

McKinsey & Company

McKinsey Global Institute | Generative AI and the future of work in America v

© Nitat Termmee / Getty
 Executive summary

The US labor market is going through a rapid evolution in the way people work and the work
people do. Months after MGI released its last report on the future of work in America, the world
found itself battling a global pandemic.1 Since then, the US job market has come roaring back
from its sudden drop. The nature of work has changed as many workers have stuck with remote
or hybrid models and employers have sped up their adoption of automation technologies. More
recently, the accelerated development of generative AI, with its advanced natural language
capabilities, has extended the possibilities for automation to a much wider set of occupations.
Amid this disruption, workers changed jobs at a remarkable pace—and a subset made bigger
leaps and moved into entirely different occupations. Some 8.6 million occupational shifts took
place from 2019 through 2022. Now even more change is in store. We expect an additional
12 million occupational shifts by 2030. The total number of transitions through 2030 could be
25 percent higher than we projected a little over two years ago.2
Multiple forces are set to fuel growth in certain occupations and erode jobs in others. They
generally fall into three categories: automation, including generative AI; an injection of federal
investment into infrastructure and the net-zero transition; and long-term structural trends such
as aging, continuing investment in technology, and the growth of e-commerce and remote work.
We do not forecast how aggregated employment may be affected by the business cycle in the
short term; instead, we focus on how these forces may reshape the composition of labor demand
over the long term.
Across a majority of occupations (employing 75 percent of the workforce), the pandemic
accelerated trends that could persist through the end of the decade. Occupations that took a hit
during the downturn are likely to continue shrinking over time. These include customer-facing
roles affected by the shift to e-commerce and office support roles that could be eliminated
either by automation or by fewer people coming into physical offices. Declines in food services,
customer service and sales, office support, and production work could account for almost
ten million (more than 84 percent) of the 12 million occupational shifts expected by 2030.

Multiple forces are set to fuel

growth in certain occupations
and erode jobs in others.

1
The future of work in America: People and places, today and tomorrow, McKinsey Global Institute, July 2019.
2
The future of work after COVID-19, McKinsey Global Institute, February 2021.

McKinsey Global Institute | Generative AI and the future of work in America 1

 By contrast, occupations in business and legal professions, management, healthcare,
transportation, and STEM were resilient during the pandemic and are poised for continued
growth. These categories are expected to see fewer than one million occupational shifts
by 2030.
For the other categories that account for the remaining one million occupational shifts still to
come, the pandemic was a temporary headwind. Employment in fields like education and training
should rise in the years ahead amid a continuous need for early education and lifelong learning.
Demand for construction workers also stalled during the height of the pandemic but is expected
to rebound strongly.
The changes estimated in our earlier research are happening even faster and on an even bigger
scale than expected. It is becoming even more urgent to solve occupational and geographic
mismatches and connect workers with the training they need to land jobs with better prospects.
The fact that workers have been willing to pivot and change career paths, while a tighter labor
market encouraged companies to hire from broader applicant pools, gives cause for optimism—
but not complacency. The future of work is already here, and it’s moving fast.

In a tighter labor market, workers have been moving

into new roles, accelerating occupational shifts
By the end of 2022, employment had bounced back to its 2019 level. But a great deal was in flux.
Are pandemic-era labor shortages here to stay?
The quits rate soared to new heights during the pandemic, with roughly 48 million Americans
leaving their jobs in 2021 and 51 million in 2022. What people did next is not fully evident from
the data. Some moved into better jobs with higher pay. Others left the labor force, whether out of
discouragement or for personal or health reasons, and it is unclear if or when they will return.
Total employment hit an all-time high after the pandemic, with many employers encountering
hiring difficulties. As of April 2023, some ten million positions remained vacant; labor force
participation had ticked up but was 0.7 percentage point below its prepandemic level. That
translates into roughly 1.9 million workers who are neither employed nor actively looking for jobs.
This erosion comes after an extended 20-year trend of steadily falling participation.
Labor supply may continue to be constrained, given that one in four Americans will be of
retirement age or older by 2030. Without higher participation rates, increased immigration,
or meaningful productivity growth, labor shortages could be a lasting issue as the economy
and the population grow. This remains an open question confronting markets, economists,
and employers.

Workers have shown a willingness

to change career paths, while
a tighter labor market has
encouraged companies to hire
from broader applicant pools.
McKinsey Global Institute | Generative AI and the future of work in America 2
Web 2023
future-of-work
Exhibit 2 and 7 of 21

Exhibit E1
More than 50 percent of recent occupational shifts in the United States involved workers
leaving roles in food services, customer service, office support, and production.
Estimated shifts to another occupation,
by category,¹ 2019–22, % (XX) — Number of occupational shifts in each
occupational category, 2019–22

Food services (1.3M) Health aides, technicians,

and wellness (700K)
>75% low-wage jobs
>75% low-wage jobs
>75% workers without college degree Hit and Resilient
Number of shifts declining Health aides
and growing >75% workers without college degree

Top 3 occupations over 2019–22 occupations Food

services Business occupations Nursing assistants 93K
Fast food and counter workers 529K Hit during COVID-19 and legal Resilient during COVID-19 Recreation workers 87K
Waiters and waitresses 397K and continuing to 8 professionals and continuing to grow Childcare workers 85K
Cooks 96K decline 16
6 Business and legal professionals (600K)
STEM
professionals <25% low-wage jobs
Customer service and sales (1.3M)
5 25–50% workers without college degree
>75% low-wage jobs
>75% workers without college degree ~8.6M total Project management specialists
Sales representatives
110K
100K
occupational
Customer 7 Others
Retail salespersons 447K 15
service and Business operations specialists 38K
Cashiers 158K sales
Hairdressers, hairstylists,
and cosmetologists
96K
shifts 5
STEM professionals (400K)
50% faster rate of change <25% low-wage jobs
Educators
Office support (1.2M) than in previous 3 years 25–50% workers without college degree
4
>75% low-wage jobs Computer systems analysts 66K
Builders
>70% workers without college degree 14 3 Computer programmers 56K

Office clerks, general 443K Electrical and electronic engineering 21K

Office 7 Community technologists and technicians
Secretaries and administrative 96K support services
assistants 10
Others (600K)
First-line supervisors of office and 70K Others
administrative support workers <25% low-wage jobs
Production
work Stalled 25–50% workers without college degree

Production work (900K) but rising Light truck drivers 62K

>75% low-wage jobs occupations Bus drivers, transit and intercity 35K
>75% workers without college degree Stalled during COVID-19 School psychologists 25K
but starting to rise Other categories include health professionals, managers, and
Laborers and freight, stock, 126K transportation services.
and material movers
Production helpers 68K
Machinists 66K
Education and workforce training (400K) Builders (300K) Community services (300K) Others (600K)

25–50% low-wage jobs 25–50% low-wage jobs 25–50% low-wage jobs 50–75% low-wage jobs
<25% workers without college degree >75% workers without college degree 50–75% workers without college degree >75% workers without college degree
Note: Figures may not sum to 100%, due to rounding.
1“Occupational shifts” refers to net declines in employment in specific occupations Substitute teachers 154K Carpenters 40K Correctional officers and jailers 65K Maids and housekeeping cleaners 134K
between 2019 and 2022. However, we do not know exactly how individuals moved from
one occupation to another or if they made multiple moves; for that reason, we refer to the Tutors 81K Painters, construction and maintenance 25K Lifeguards, ski patrol, and 36K Coaches and scouts 26K
number of occupational shifts rather than specifying the number of workers making those other recreational protective
changes. Preschool teachers 25K Drywall and ceiling tile installers 14K Computer, automated teller, 23K
Source: O*NET; US Bureau of Labor Statistics; Current Population Survey, US Census service workers
and office machine repairers
Bureau; McKinsey Global Institute analysis
Rehabilitation counselors 25K Other categories include agriculture, creatives and art
management, mechanical installation and repair, and
McKinsey & Company property maintenance.

McKinsey Global Institute | Generative AI and the future of work in America 3

The Great Attrition obscured deeper shifts
While most attention was focused on soaring quits rates during the pandemic, something more
structural was also occurring. A subset of people did more than change employers; they moved
into different occupations altogether. Based on net increases and decreases in employment,
some 8.6 million occupational shifts took place from 2019 through 2022—50 percent more than
in the previous three-year period (Exhibit E1). 3 While it is impossible to trace individual moves,
many people left their previous roles and landed better-paying jobs in other occupations.
The majority of these shifts came from people leaving jobs in food services, customer service and
sales, office support, and production work (such as manufacturing). At the same time, managerial
and professional roles plus transportation services collectively added close to four million jobs
from 2019 to 2022. Our previous research had anticipated these types of changes over a longer
time frame, but the pandemic suddenly accelerated matters. The past few years have been a
preview of trends we expect to continue through the end of the decade.
More high-wage jobs—and fewer workers taking lower-wage service jobs
Overall employment in low- and middle-wage occupations has fallen from prepandemic levels,
while occupations that pay more than $57,000 annually added about 3.5 million jobs. However, it
is unclear how many higher-paying roles were filled by people who moved up and how many were
filled by new entrants to the labor force. Meanwhile, the number of lower-wage job openings has
not declined. Demand for lower-wage service work remains, but fewer workers are accepting
these roles.
What is clear from the job switching and occupational shifts of the past three years is that the
US labor market accommodated a higher level of dynamic movement. Spiking demand and
labor scarcity forced many employers to consider nontraditional candidates with potential and
train them if they lacked direct experience. While this may not hold in the future, employers and
workers alike can draw on what they have learned about the potential for people to make quick
pivots and add new skills.

Automation and other forces will continue to reshape the labor market
Automation, from industrial robots to automated document processing systems, continues to
be the biggest factor in changing the demand for various occupations. Generative AI is both
accelerating automation and extending it to an entirely new set of occupations. While this
technology is advancing rapidly, other forces are also affecting labor demand. Overall, we expect
significant shifts in the occupational mix in the United States through the end of the decade.
The effects of automation and generative AI
Automation has taken a leap forward with the recent introduction of generative AI tools.
“Generative” refers to the fact that these tools can identify patterns across enormous sets of data
and generate new content—an ability that has often been considered uniquely human. Their most
striking advance is in natural language capabilities, which are required for a large number of work
activities. While ChatGPT is focused on text, other AI systems from major platforms can generate
images, video, and audio.
Although generative AI is still in the early stages, the potential applications for businesses are
significant and wide-ranging. Generative AI can be used to write code, design products, create
marketing content and strategies, streamline operations, analyze legal documents, provide
customer service via chatbots, and even accelerate scientific discovery. It can be used on its own
or with “humans in the loop”; the latter is more likely at present, given its current level of maturity.

3
Measured as net job losses for individual occupations across sectors, net of estimated retirements; derived from US Bureau
of Labor Statistics (BLS) data. An administrative assistant who takes a similar position with another employer has simply
switched jobs and is not part of this analysis. If that person becomes an office manager, they have changed occupations
within the same category (office support). If they become a computer systems analyst, they have moved into a different
occupational category (STEM professionals). The latter two moves are the kind of occupational shifts that we measure. Since
we are unable to trace exactly how individual workers moved, we use net declines as a broad proxy. In our forward-looking
scenario, we refer to people needing to make transitions if demand is projected to decline in their current occupation.

McKinsey Global Institute | Generative AI and the future of work in America 4

 All of this means that automation is about to affect a wider set of work activities involving
expertise, interaction with people, and creativity. The timeline for automation adoption could be
sharply accelerated. Without generative AI, our research estimated, automation could take over
tasks accounting for 21.5 percent of the hours worked in the US economy by 2030. With it, that
share has now jumped to 29.5 percent (Exhibit E2). 4

4
Note that this is the midpoint, representing the average of a very wide range, from 3.7 to 55.3 percent.
Web 2023
future-of-work
Exhibit 3 and 13 of 21
Exhibit E2
With generative AI added to the picture, 30 percent of hours worked today
could be automated by 2030.

Midpoint automation adoption¹ by 2030 as a share of time spent on work activities, US, %

Automation adoption without Automation adoption with XX — Percentage-point acceleration in

generative AI acceleration generative AI acceleration automation adoption from generative AI

0 10 20 30 40

STEM professionals 16
Education and workforce training 16
Creatives and arts management 15
Business and legal professionals 14
Managers 9
Community services 9
Office support 7
Health professionals 6
Builders 6
Property maintenance 6
Customer service and sales 6
Food services 5
Transportation services 5
Mechanical installation and repair 5
Production work 4
Health aides, technicians, and wellness 4
Agriculture 3
All sectors² 8

¹Midpoint automation adoption is the average of early and late automation adoption scenarios as referenced in The economic potential of generative AI: The next
productivity frontier, McKinsey & Company, June 2023.
²Totals are weighted by 2022 employment in each occupation.
Source: O*NET; US Bureau of Labor Statistics; McKinsey Global Institute analysis

McKinsey & Company

McKinsey Global Institute | Generative AI and the future of work in America 5
Other forces affecting future labor demand
Automation is not occurring in a vacuum, of course. Other trends are affecting the demand for
certain occupations, and we expect the employment mix to change significantly through 2030,
with more healthcare, STEM, and managerial positions and fewer jobs in customer service, office
support, and food services.
— Federal investment: Recent federal legislation is driving momentum and investment in
other areas that will affect jobs. 5 Reaching the net-zero emissions goal is one of these
priorities. Some 3.5 million jobs could be displaced through direct and indirect effects across
the economy. But at the macro level, these losses should be more than offset by gains of
4.2 million jobs, primarily led by capital expenditures on renewable energy. The net-zero
transition will likely be a net positive for jobs, but those jobs may be located in different places
and require different skills.
Similarly, major investment in infrastructure projects across the country will bolster
construction jobs, which could see employment growth of 12 percent from 2022 through
2030. However, the sector already had some 383,000 unfilled positions in April 2023. This
shortage will have to be addressed to bring infrastructure projects to life from coast to coast.6
The CHIPS and Science Act is putting additional funding into semiconductor manufacturing
as well as R&D and scientific research.7 This comes at a time when some companies have
been adjusting their supply chains, leading to an uptick in domestic manufacturing. While
manufacturing is likely to boost employment demand overall in the years ahead, the sector is
becoming more high-tech. It will involve fewer traditional production jobs than in the past but
more workers with technical and STEM skills. 8
— Other structural trends: At the same time, other trends like rising incomes and education
levels will sustain jobs. An aging population will need more healthcare workers in multiple
roles, while the ongoing process of digitizing the economy will require adding more tech
workers in every sector.

Putting it all together, the mix of jobs is changing, and we

anticipate an additional 12 million occupational shifts
One of the biggest questions of recent months is whether generative AI might wipe out jobs. Our
research does not lead us to that conclusion, although we cannot definitively rule out job losses,
at least in the short term. Technological advances often cause disruption, but historically, they
eventually fuel economic and employment growth.
This research does not predict aggregated future employment levels; instead, we model various
drivers of labor demand to look at how the mix of jobs might change—and those results yield
some gains and some losses.9 In fact, the occupational categories most exposed to generative
AI could continue to add jobs through 2030 (Exhibit E3), although its adoption may slow their
rate of growth. And even as automation takes hold, investment and structural drivers will support
employment. The biggest impact for knowledge workers that we can state with certainty is that
generative AI is likely to significantly change their mix of work activities.

5
While our scenario includes the impact of federal investment in the net-zero transition and infrastructure, it does not include
the full impact of the CHIPS and Science Act and the Inflation Reduction Act, since implementation remained unclear at the
time of this analysis. However, both pieces of legislation point to the possibility of additional upside.
6
Garo Hovnanian, Adi Kumar, and Ryan Luby, “Will a labor crunch derail plans to upgrade US infrastructure?” McKinsey &
Company, October 2022.
7
Note that both the CHIPS and Science Act and the Inflation Reduction Act create room for additional upside in employment.
But since there is still uncertainty about their implementation as of this writing, their effects on jobs are not explicitly
incorporated into our scenario.
8
For more on this topic, see Asutosh Padhi, Gaurav Batra, and Nick Santhanam, The titanium economy: How industrial
technology can create a better, faster, stronger America, Public Affairs, 2022.
9
We rely on employment projections from the US Bureau of Labor Statistics for 2030 employment levels.

McKinsey Global Institute | Generative AI and the future of work in America 6

Web 2023
future-of-work
Exhibit E3 8 of 21
Exhibit 4 and

While STEM, healthcare, builders, and professional fields continue to add jobs,
generative AI could change work activities significantly for many occupations.

Estimated labor demand change and generative AI Midpoint automation Employment,

automation acceleration by occupation, US, 2022–30 adoption¹ by 2030, % absolute
15– 25– 35–
35 25 35 40 5M 10M
Health
professionals

30

Health aides,
technicians,
25 and wellness
STEM
professionals

20 Increasing labor demand Increasing labor demand

and modest change of and high change of work
work activities activities

15
Builders Managers Creatives and
Change arts management
in labor
demand,² % 10 Transportation
Property
services
maintenance
Business and
legal professionals
5 Mechanical Community
Agriculture installation services Education and
and repair workforce
training
0
Production 5 10 15 20
work Food
services
–5

Decreasing labor demand

with modest change of
work activities
–10

Customer
service
–15 and sales

Office
support
–20

Increase in automation adoption driven by generative AI acceleration, percentage points

¹Midpoint automation adoption is the average of early and late automation adoption scenarios as referenced in The economic potential of generative AI: The next
productivity frontier, McKinsey & Company, June 2023.
2We consider multiple drivers affecting demand: rising income, aging populations, technology investment, infrastructure investment (including Bipartisan
Infrastructure Law), rising education levels, net-zero transitions, marketization of unpaid work, creation of new occupations, automation (including generative AI),
increased remote working and virtual meetings, and e-commerce and other virtual transactions.
Source: US Bureau of Labor Statistics; Current Population Survey, US Census Bureau; McKinsey Global Institute analysis

McKinsey & Company

McKinsey Global Institute | Generative AI and the future of work in America 7

 Resilient and growing occupational categories
The largest future job gains are expected to be in healthcare, an industry that already has an
imbalance, with 1.9 million unfilled openings as of April 2023. We estimate that there could be
demand for 3.5 million more jobs for health aides, health technicians, and wellness workers, plus
an additional two million healthcare professionals.10
By 2030, we further estimate a 23 percent increase in the demand for STEM jobs. Although
layoffs in the tech sector have been making headlines in 2023, this does not change the longer-
term demand for tech talent among companies of all sizes and sectors as the economy continues
to digitize. Employers in banking, insurance, pharmaceuticals, and healthcare, for example,
are undertaking major digital transformations and need tech workers with advanced skills.11
In addition, the transportation services category is expected to see job growth of 9 percent
by 2030.
Declining occupational categories
The biggest future job losses are likely to occur in office support, customer service, and food
services. We estimate that demand for clerks12 could decrease by 1.6 million jobs, in addition to
losses of 830,000 for retail salespersons, 710,000 for administrative assistants, and 630,000
for cashiers. These jobs involve a high share of repetitive tasks, data collection, and elementary
data processing, all activities that automated systems can handle efficiently. Our analysis also
finds a modest decline in production jobs despite an upswing in the overall US manufacturing
sector, which is explained by the fact that the sector increasingly requires fewer traditional
production jobs but more skilled technical and digital roles.13
We estimate that 11.8 million workers currently in occupations with shrinking demand may need
to move into different lines of work by 2030. Roughly nine million of them may wind up moving
into different occupational categories altogether. Considering what has already transpired, that
would bring the total number of occupational transitions through the decade’s end to a level
almost 25 percent higher than our earlier estimates, creating a more pronounced shift in the mix
of jobs across the economy.
Overall, we expect more growth in demand for jobs requiring higher levels of education and skills,
plus declines in roles that typically do not require college degrees (Exhibit E4).

Almost 12 million additional

occupational transitions may be
needed by the end of the decade.

10
Note that registered nurses, nurse practitioners, and nurse anesthetists are in the healthcare professionals category; nurse
midwives and licensed practical and licensed vocational nurses are in the health aides category.
11
Jon Swartz, “As Big Tech cuts workers, other industries are desperate to hire them,” MarketWatch, February 18, 2023; and
Steve Lohr and Tripp Mickle, “As Silicon Valley retrenches, a tech talent shift accelerates,” New York Times, December 29,
2022.
12
Note that clerks include receptionists and information clerks, general office clerks, bookkeeping, accounting, and auditing
clerks, and shipping, receiving, and inventory clerks
13
Building a more competitive US manufacturing sector, McKinsey Global Institute, April 2021.

McKinsey Global Institute | Generative AI and the future of work in America 8

Web 2023
future-of-work
Exhibit 5 of
Exhibit E421

Healthcare, STEM, and builder roles could grow, while demand for office
support and customer service roles could decline.

Estimated future US job growth by occupational category

Midpoint automation scenario,¹ with generative AI acceleration
Resilient and growing occupations² Stalled but rising occupations³ Hit and declining occupations⁴
Employment,
Occupational category Net change in labor demand, 2022–30, % 2022, million
Health professionals 30 6.5

Health aides, technicians, and wellness 30 11.6

STEM professionals 23 7.9

Builders 12 7.0

Managers 11 9.7

Creatives and arts management 11 2.2

Property maintenance 10 4.6
Transportation services 9 5.6
Mechanical installation and repair 7 6.6
Business and legal professionals 7 16.0
Community services 7 6.8
Education and workforce training 3 9.9
Agriculture 2 2.1

Production work –1 13.3

Food services –2 13.7

Customer service and sales –13 14.7

Office support –18 20.1

¹Midpoint automation adoption is the average of early and late automation adoption scenarios as referenced in The economic potential of generative AI: The next
productivity frontier, McKinsey & Company, June 2023.
2Resilient during the pandemic, 2019–22, and expected to grow between 2022 and 2030.
3Stalled during the pandemic, 2019–22, and expected to rise between 2022 and 2030.
4Hit during the pandemic, 2019–22, and continuing to decline between 2022 and 2030.
Source: O*NET; US Bureau of Labor Statistics; Current Population Survey, US Census Bureau; McKinsey Global Institute analysis

McKinsey & Company

Workers in jobs with lower wages and educational
requirements could be the most affected
People in the two lowest wage quintiles (those earning less than $30,800 a year and those
earning $30,800 to $38,200 a year) are up to 10 and 14 times more likely, respectively, to need to
change occupations by the end of this decade than the highest earners.14 Changing occupations,
as opposed to finding a new job within the same occupation, often requires adding new skills and
is more challenging.

14
Workers in the second-lowest wage quintile (earning $30,800 to $38,200 per year) have an even higher risk of needing
to transition occupations than the lowest-wage workers, who earn less than $30,800 a year. Replacing the lowest-wage
workers with technology may not make economic sense, but at a certain wage level, the equation changes. In addition, some
lower-wage jobs involve unpredictable physical work or customer-facing work that does not lend itself well to automation.

McKinsey Global Institute | Generative AI and the future of work in America 9

The jobs in the two lowest wage quintiles are disproportionately held today by those with less
education, women, and people of color. Women are heavily represented in office support and
customer service, which could shrink by about 3.7 million and 2.0 million jobs, respectively,
by 2030. Similarly, Black and Hispanic workers are highly concentrated in some shrinking
occupations within customer service, food services, and production work.
While our analysis shows a decrease of 1.1 million jobs in the two lowest wage quintiles by
2030, jobs in the highest wage quintile could grow sharply, by 3.8 million. Helping workers in
lower-wage, shrinking occupations move into better-paying jobs with more stability will require
widespread access to training programs, effective job matching, different hiring and training
practices by employers, and better geographic mobility.
The overall labor market will have higher demand for social-emotional and digital skills. Although
the demand for basic cognitive and manual skills is likely to decline, physical work is not going
away. It may still account for just under 31 percent of time spent, driven by growth in sectors such
as transportation services, construction, and healthcare.

This period of change can be an opportunity for more inclusive growth

With the pace of change unlikely to let up, the challenge will be helping workers match up with the
jobs of the future. While some of this may require large-scale collaboration, individual companies
can fill many of the gaps by adapting their own approaches to hiring and training.
Boosting productivity through automation and generative AI
Recent MGI research focused on how to reignite productivity growth in the United States.15
Automation and reskilling will be vital to this effort. Automation could jump-start lackluster
productivity while simultaneously easing labor shortages.
Generative AI has the potential to increase US labor productivity by 0.5 to 0.9 percentage
points annually through 2030 in a midpoint adoption scenario. The range reflects whether the
time freed up by automation is redeployed at 2022 productivity levels or 2030 levels, with both
scenarios accounting for the occupational mix expected in 2030.
Combining generative AI with all other automation technologies, the potential growth could be
even larger. All types of automation could help drive US productivity growth to 3 to 4 percent
annually in a midpoint adoption scenario. However, this will require significant action from
stakeholders across the public and private sector. Workers will need support in learning new
skills, and other risks associated with generative AI also need to be mitigated and controlled. But
if worker transitions and risks are well managed, generative AI could contribute substantively to
economic growth.
To capture the full benefits of generative AI to make knowledge work more productive,
employers, policy makers, and broader ecosystems would need to establish clear guidelines
and guardrails—and workers would need to see these tools not as job destroyers but as work
enhancers. When machines take over dull or unpleasant tasks, people can be left with more
interesting work that requires creativity, problem-solving, and collaborating with others. Workers
will need to gain proficiency with these tools and, importantly, use the time that is freed up to
focus on higher-value activities. When managers automate more of their administrative and
reporting tasks, for example, they can spend more time on strategic thinking and coaching.
Similarly, researchers could speed up projects by relying on automation tools to sort and
synthesize large data sets.
For employers, doubling down on innovative hiring strategies
Most employers can benefit from using a broader lens in hiring. Instead of insisting on prior
experience that matches the responsibilities of an open role as closely as possible, organizations
can evaluate candidates on their capacity to learn, their intrinsic capabilities, and their
transferable skills.

15
Rekindling US productivity for a new era, McKinsey Global Institute, February 2023.

McKinsey Global Institute | Generative AI and the future of work in America 10

A great deal of skills development happens on the job. Previous MGI research found that work
experience contributes 40 percent of the average individual’s lifetime earnings in the United
States.16 Skills learned through work experience are an even bigger determinant for people
without educational credentials who start out in lower-wage work.
Filling the jobs of the future is an opportunity to make the labor market more inclusive. Employers
may need to reconsider whether some credential requirements are really necessary. Some
60 percent of US workers have skills gained through experience but lack four-year college
degrees. Initiatives like Tear the Paper Ceiling are supporting workers who have experience but
not degrees by raising awareness among employers and providing resources.
Employers can also recruit from populations that are often overlooked, such as retirees who want
to return to work, people with employment gaps, and the formerly incarcerated. Remote work,
for example, is opening up long-needed opportunities for people with disabilities who cannot
commute and those in rural communities.
Tackling other structural issues
Women left the workforce in relatively higher numbers than men during the pandemic. It took
three full years for the number of working women in the United States to fully bounce back. Many
women doing lower-wage work have family obligations that may leave them feeling that they can’t
take the risk of going back to school or trying a new occupation. Beyond the hiring practices that
can encourage and enable women to make career transitions, the need for affordable childcare
remains a major barrier.17 To address it, a number of private-sector employers are expanding
childcare benefits, while some state and local governments are providing tax credits, subsidies,
or direct funding. In addition, historically male-dominated fields such as construction that are
facing labor shortages can fill those gaps with more women, improving diversity in the process.
One key area of job demand is in caregiving, which is critical social infrastructure. We anticipate
that the two fastest-growing occupations through the end of this decade will be nurses and home
healthcare aides.18 Childcare workers, as noted above, provide a vital service to working families.
But people have been leaving these types of jobs in droves. Meeting these growing needs
will likely hinge on upgrading the quality of what are today typically low-paying jobs with little
security or advancement opportunities.
While large employers may be able to handle their own training needs, the magnitude of the
reskilling challenge for the nation as a whole calls for broader partnerships with industry groups,
educational providers, and nonprofits as well as incentives for investing in human capital.
Addressing the need for reskilling with efforts beyond individual companies would help spread
the cost, addressing the concerns of employers who might be reluctant to invest in training for
employees who can subsequently leave.
With millions of jobs potentially being eliminated by automation—and even more being created
in fields requiring different skills—the United States needs broad access to effective training
programs as well as job-matching assistance that can help individuals find opportunities.
Many initiatives are in place, but it will be critical to dramatically scale up what works and take
a proactive approach to filling key shortages. One promising solution, still in the early stages,
involves digital learning and employment records—a kind of digital microcredential that can
document how an individual worker has acquired skills and also translate across companies and
over time.
The US labor market has been remarkably resilient in the face of recent challenges and rapid
changes. That kind of adaptability is exactly what it will take to navigate the next chapter as well,
supporting individuals while helping businesses meet their talent needs so they can continue
driving growth.

16
Human capital at work: The value of experience, McKinsey Global Institute, June 2022.
17
Sarah Gitlin, Ayushi Gummadi, Alexis Krivkovich, and Kunal Modi, “The childcare conundrum: How can companies ease
working parents’ return to the office?” McKinsey & Company, May 2022.
18
For occupations that employed more than 50,000 people as of 2022.

McKinsey Global Institute | Generative AI and the future of work in America 11

© Goroden Koff/Getty
Introduction

This report offers a fresh look at how the future of work could shape up for the United States.
It’s a timely moment to take stock; the landscape has changed dramatically since the McKinsey
Global Institute published research on automation and the US workplace in 2019. We found that
the trends we were anticipating in our previous research have indeed been unfolding—and even
faster than our research originally indicated.
The US labor market had just come out of the pandemic when a revolutionary technology
development burst on the scene. ChatGPT, along with other generative AI tools, attracted
millions of users virtually overnight. The potential applications for cognitive and creative tasks
have raised questions about the implications for workers. This research attempts to answer those
questions while putting generative AI in context with many other trends that are likely to raise or
lower demand for certain occupations.
Our forward-looking analysis incorporates the pandemic-era changes that appear to be lasting,
along with new estimates for technology adoption and advancement. This time, it also considers
trends fueled by major infusions of federal investment in the clean-energy transition and the
renewal of the nation’s infrastructure.19 Our primary focus is on labor demand, although we
include some observations about the future labor supply and how technology might change the
day-to-day nature of work.
We look to the future with full awareness of how much is uncertain. Some data on what is
occurring today is still limited, and future disruptions could always change the trajectory. This
research is not meant to provide a short-term job forecast affected by fluctuations in the
business cycle. Instead, it offers a longer-term structural view of how the US labor market could
keep evolving.
The clear takeaway is the need to prepare for continued big occupational shifts in the years
ahead. Millions of jobs could be phased out even as new ones are created. Preparing American
workers for the jobs of the future is a top priority for leaders across the private, public, and social
sectors. If handled well, this period of change could create a more inclusive economy with higher
productivity growth.
The future of work is no longer something to look for over the horizon. We’re already in the midst
of it.

19
While our scenario includes the impact of federal investment in the net-zero transition and infrastructure, it does not include
the full impact of the CHIPS and Science Act and the Inflation Reduction Act, since implementation remained unclear at the
time of this analysis. However, both pieces of legislation point to the possibility of additional upside.

McKinsey Global Institute | Generative AI and the future of work in America 13

© PM Images /Getty
1. A robust recovery
marked by job
switching and
labor shortages

Gaining a clearer picture of how America’s future of work could shape up through the end of this
decade involves taking the nation’s new starting point into account. More than two years after
the initial COVID-19 shutdowns, the US labor market has regained its footing, with the economy
continuing to add jobs steadily. The unemployment rate dipped below 4 percent at the beginning
of 2022 and has remained there through the first half of 2023.
Against the backdrop of a tight labor market, a massive wave of job switching occurred. About
48 million people quit their jobs in 2021, followed by 51 million in 2022. With openings readily
available, many workers felt empowered to seek out new positions that offered higher pay or
better alignment with priorities such as flexibility or opportunities to advance.20
A subset of those people did more than simply change employers while doing essentially the
same job. They moved into different occupations altogether, accelerating shifts in the mix of
employment that our ongoing body of research has been anticipating. In this chapter, we take
stock of these recent changes, which inform the updated baseline and reaffirm some of the
drivers of labor demand used in our refreshed research.

The pandemic changed the mix of employment, with

tougher implications for workers with less education
From the very start of the pandemic, it was evident that demand would rachet up for some types
of work and dry up for others, simply due to which occupations were deemed essential or could
be performed remotely. This unevenness has held true throughout the recovery (Exhibit 1).21
These changes offer a preview of trends that we expect to continue through the end of
the decade.
Managerial roles and business, legal, and STEM professions were resilient during the pandemic,
since these jobs could be done remotely and were often critical to helping businesses navigate
new challenges. These categories added 3.3 million jobs from 2019 to 2022. At the same time,
e-commerce fueled growth in transportation services—specifically in warehousing. Jobs for
stockers and order fillers, for example, grew by more than 30 percent between 2019 to 2022,
while shipping, receiving, and inventory clerks grew by about 15 percent. These trends are likely
to continue.

20
See, for example, Jo Constanz, “The Great Resignation worked: Most job-swappers got a raise,” Bloomberg, July 29, 2022;
and Te-Ping Chen, “Better pay and career paths drive US workers’ decisions to quit,” Wall Street Journal, March 9, 2022.
21
As of this writing, 2022 is the latest year with detailed data on employment by occupation; see “Occupational Employment
and Wage Statistics,” US Bureau of Labor Statistics.

McKinsey Global Institute | Generative AI and the future of work in America 15

Web 2023
future-of-work
Exhibit
Exhibit 6 of121

Customer-facing jobs sustained the biggest losses due to the pandemic, while
business, transportation, and STEM roles were more resilient.

Employment change by occupational category, 2019–22

Resilient and growing occupations¹ Stalled but rising occupations² Hit and declining occupations³

Employment change, Employment change, Employment,

Occupational category 2019–22, absolute 2019–22, % 2019, million
Managers 1.6M 20 8.1
Business and legal professionals 1.2M 8 14.8
STEM professionals 509K 7 7.4
Health professionals 249K 4 6.3
Transportation services 224K 4 5.4
Mechanical installation and repair 81K 1 6.5
Agriculture 30K 1 2.1
Creatives and arts management 8K 0 2.2
Community services –82K –1 6.9
Builders –113K –2 7.1
Property maintenance –124K –3 4.7
Production work –128K –1 13.4
Office support –128K –1 20.3
Health aides, technicians, and wellness –163K –1 11.8
Education and workforce training –298K –3 10.2
Food services –844K –6 14.6
Customer service and sales –1.3M –8 16.0

1Resilient during the pandemic, 2019–22, and expected to grow between 2022 and 2030.
2Stalled during the pandemic, 2019–22, and expected to rise between 2022 and 2030.
3Hit during the pandemic, 2019–22, and continuing to decline between 2022 and 2030.
Source: O*NET; US Bureau of Labor Statistics; Current Population Survey, US Census Bureau; McKinsey Global Institute analysis

McKinsey & Company

Most other occupational groups did not recover so quickly, and the categories that took the
biggest hits are likely to continue declining. During the height of the pandemic, many food service
jobs were lost. Once businesses reopened their doors, however, they struggled to hire and retain
workers. While total employment in the leisure and hospitality sector remained down throughout
2021, it has largely recovered. But as of April 2023, employment specifically in accommodation
and food services was still slightly below prepandemic levels, with about 40 percent more job
openings than before the pandemic.
Customer service and sales roles decreased throughout the pandemic as many brick-and-
mortar retail locations closed. They have recovered slowly; as of 2022, customer service
employment was still 8 percent below its 2019 level, which equates to 1.3 million fewer jobs. Much
of this is due to consumers’ changing preferences for e-commerce.22
Office support remained 1 percent below its 2019 levels as of 2022. In a recent McKinsey
survey, nearly half of respondents said the COVID-19 pandemic accelerated their organization’s
deployment of automation technologies such as intelligent document management and
processing tools.23 Office clerks, secretaries, and tellers were most affected, a trend that is likely
to continue as automation continues to become more ubiquitous.
The recovery in healthcare employment is more nuanced. The overall healthcare sector was
down about 400,000 jobs from 2019 to 2021, although it has since recovered. While employment
increased for health professionals, it has continued to lag for health aides, technicians, and
wellness workers. The issue has not been lack of demand but rather lack of workers. This is likely

22
The future of work after COVID-19, McKinsey Global Institute, February 2021.
23
“Your questions about automation, answered,” McKinsey Global Survey, McKinsey & Company, July 2022.

McKinsey Global Institute | Generative AI and the future of work in America 16

due to a combination of burnout, particularly for those in frontline care roles, and workers in the
lowest-paid roles leaving the field.24 There is potential for shortages to continue unless these
roles can become more attractive for workers.
Mirroring the uneven gains and losses across occupations were uneven effects for different
segments of the labor force. Those without four-year college degrees were harder hit than
college-educated workers. People of color disproportionately struggled since they were
overrepresented in occupations that sustained losses. While the national unemployment
rate during COVID-19 peaked at 14.7 percent in April 2020, it hit 18.8 percent for Hispanic
workers in the same month and 16.9 percent for Black workers a month later. Women were also
concentrated in sectors such as accommodation and education, which shed about 700,000 and
300,000 jobs, respectively, between 2019 and 2022.

In the economic recovery, the United States developed a labor shortage

As of May 2023, total employment stood at two million above its prepandemic level. While labor
force participation has ticked back up, it remained 0.7 percentage point below where it was
in February 2020. That translates into roughly 1.9 million fewer workers at a time when about
ten million positions remained vacant. The job market was also tight in 2018 and 2019, but the
postpandemic labor shortage has been even more pronounced. In fact, taking a longer historical
view, labor force participation has been steadily falling, from 67.3 percent in January 2000 to
62.6 percent in May 2023.
Despite a wave of white-collar layoffs in 2023 that received substantial media coverage, the job
market has remained robust overall, especially for blue-collar workers.25 Many employers have
been coping with unfilled positions and competing for talent. This has driven wages upward, in
stark contrast to the prolonged weak jobs recovery that followed the Great Recession of 2008.
Where are the workers?
A great deal of uncertainty surrounds the current labor shortage. It is unclear whether people
have left the labor force permanently or might return. Several drivers help to explain at least part
of what is going on and shed light on who might be drawn back in to meet future demand and fill
more of the ten million positions that are vacant today as well as those that will be created at a
time when the workforce is shrinking due to aging.26
— People have retired. According to the US Bureau of Labor Statistics (BLS), labor force
participation for Americans over age 65 fell from 20.8 percent in February 2020 to
19.1 percent in May 2023. Additionally, some workers in their late 50s and early 60s exited
the workforce early. One study estimates that half of the recent increase in retirements was
driven by “excess retirements” above the prepandemic trend line. This translates into roughly
1.6 million missing older workers.27
— People have quit. During the pandemic, an unprecedented number of workers headed for
the exits, roughly a third of whom left without having their next jobs lined up.28 While many
people landed new positions, others appear to have left the workforce, whether temporarily
or permanently.

24
“Assessing the lingering impact of COVID-19 on the nursing workforce,” McKinsey & Company, May 2022. See also Bryan
Sexton et al., “Emotional exhaustion among US health care workers before and during the COVID-19 pandemic, 2019–2021,”
JAMA Network Open (American Medical Association), September 2022; and Carlo Giacomo Leo et al., “Burnout among
healthcare workers in the COVID-19 era: A review of the existing literature,” Frontiers in Public Health, October 2021. See also
Lauren Kaori Gurley, “Why nurses say they are striking and quitting in droves,” Washington Post, January 14, 2023.
25
Justin Lahard et al., “It’s a richcession, not a recession. Here’s your investing playbook,” Wall Street Journal, February 17,
2023.
26
“Why did the labor force participation rate decline when the economy was good?” US Census Bureau, June 2021; “Supporting
labor supply in the American Jobs Plan and the American Families Plan,” The White House, 2021; and “Can a hot but smaller
labor market keep making gains in participation?” Brookings Institution, August 2022.
27
Joshua Montes, Christopher Smith, and Juliana Dajon, “‘The great retirement boom’: The pandemic-era surge in retirements
and implications for future labor force participation,” Finance and Economics Discussion Series, Federal Reserve, November
2022.
28
Aaron De Smet, Bonnie Dowling, Marino Mugayar-Baldocchi, and Bill Schaninger, “‘Great Attrition’ or ‘Great Attraction’? The
choice is yours,” McKinsey Quarterly, September 2021.

McKinsey Global Institute | Generative AI and the future of work in America 17

— Health concerns are taking a significant if hard-to-measure toll. While data is scarce,
some people stricken with COVID-19 struggle with long-term effects that hamper their
ability to work.29 In addition, studies have documented a rise in burnout and mental
health challenges. 30
— Immigration was stalled for an extended period. Immigrants have always been part of
the American labor force. But migration into the United States began dropping in 2016, and
pandemic-related restrictions accelerated the decline further. The number of foreign-born
workers experienced a sharp decline from the onset of the pandemic. In 2016, more than
a million immigrants (including asylum seekers, students, and holders of both work and
immigrant visas) entered the country. By 2021, that number was down to fewer than 400,000.
Those annual drops compound over time. Keeping 2014–16 levels of net international
migration in place, for example, would have added 1.7 million people over 2017–21. However,
immigration is estimated to have rebounded in 2022, with net international migration
returning to 2018 levels. 31 The foreign-born share of the US labor force was higher than the
prepandemic level as of May 2023. 32
Whether current hiring difficulties turn out to be a preview of a lasting structural issue is the
question of the moment for markets, economists, and employers; it is also the subject of
forthcoming research from MGI. What we do know is that the tight environment is driving home
the urgency of ensuring that critical occupations are filled and that people have opportunities to
add the skills required to move into these roles.

Millions of workers not only changed jobs but

made bigger leaps into new occupations
The Great Attrition saw Americans switching jobs in record numbers. Within that group, a
significant subset of people not only changed employers but took up entirely new lines of work.
We estimate that some 8.6 million occupational shifts took place between 2019 and 2022—an
increase of almost 50 percent from the preceding three-year period. 33 Declines in food services,
customer service and sales, office support, and production accounted for more than half of these
transitions (Exhibit 2). Losses in these categories are part of a longer-term structural change that
we expect to continue.

29
Katie Bach, “New data shows long Covid is keeping as many as 4 million people out of work,” Brookings Institution, August
2022. This analysis is based in part on data from the US Census Bureau, the Minneapolis Fed, and The Lancet.
30
See, for example, “Present company included: Prioritizing mental health and well-being for all,” McKinsey Health Institute,
October 2022.
31
“Net migration between the United States and abroad in 2022 reaches highest level since 2017,” US Census Bureau,
December 2022.
32
Gabriel T. Rubin and Rosie Ettenheim, “Immigrants’ share of the U.S. labor force grows to a new high,” Wall Street Journal, May
22, 2023.
33
We measure occupational shifts as net job losses for individual occupations across sectors, net of estimated retirements;
derived from BLS data. An administrative assistant who leaves one employer to take a similar position with another has simply
switched jobs and is not part of this analysis. If that person becomes an office manager, they have changed occupations
within the same category (office support). If they become a computer systems analyst, they have moved into a different
occupational category (STEM professionals). The latter two moves are the kind of occupational shifts that we measure. Since
we are unable to trace exactly how individual workers moved, we use net declines as a broad proxy for how labor demand and
supply have changed across occupations. In our forward-looking scenario, we refer to people needing to make transitions if
labor demand in their current occupation is projected to decline.

McKinsey Global Institute | Generative AI and the future of work in America 18

future-of-work
Exhibit 2 and 7 of 21

Exhibit 2
More than 50 percent of recent occupational shifts in the United States involved workers
leaving roles in food services, customer service, office support, and production.
Estimated shifts to another occupation,
by category,¹ 2019–22, % (XX) — Number of occupational shifts in each
occupational category, 2019–22

Food services (1.3M) Health aides, technicians,

and wellness (700K)
>75% low-wage jobs
>75% low-wage jobs
>75% workers without college degree Hit and Resilient
Number of shifts declining Health aides
and growing >75% workers without college degree

Top 3 occupations over 2019–22 occupations Food

services Business occupations Nursing assistants 93K
Fast food and counter workers 529K Hit during COVID-19 and legal Resilient during COVID-19 Recreation workers 87K
Waiters and waitresses 397K and continuing to 8 professionals and continuing to grow Childcare workers 85K
Cooks 96K decline 16
6 Business and legal professionals (600K)
STEM
professionals <25% low-wage jobs
Customer service and sales (1.3M)
5 25–50% workers without college degree
>75% low-wage jobs
>75% workers without college degree ~8.6M total Project management specialists
Sales representatives
110K
100K
occupational
Customer 7 Others
Retail salespersons 447K 15
service and Business operations specialists 38K
Cashiers 158K sales
Hairdressers, hairstylists,
and cosmetologists
96K
shifts 5
STEM professionals (400K)
50% faster rate of change <25% low-wage jobs
Educators
Office support (1.2M) than in previous 3 years 25–50% workers without college degree
4
>75% low-wage jobs Computer systems analysts 66K
Builders
>70% workers without college degree 14 3 Computer programmers 56K

Office clerks, general 443K Electrical and electronic engineering 21K

Office 7 Community technologists and technicians
Secretaries and administrative 96K support services
assistants 10
Others (600K)
First-line supervisors of office and 70K Others
administrative support workers <25% low-wage jobs
Production
work Stalled 25–50% workers without college degree

Production work (900K) but rising Light truck drivers 62K

>75% low-wage jobs occupations Bus drivers, transit and intercity 35K
>75% workers without college degree Stalled during COVID-19 School psychologists 25K
but starting to rise Other categories include health professionals, managers, and
Laborers and freight, stock, 126K transportation services.
and material movers
Production helpers 68K
Machinists 66K
Education and workforce training (400K) Builders (300K) Community services (300K) Others (600K)

25–50% low-wage jobs 25–50% low-wage jobs 25–50% low-wage jobs 50–75% low-wage jobs
<25% workers without college degree >75% workers without college degree 50–75% workers without college degree >75% workers without college degree
Note: Figures may not sum to 100%, due to rounding.
1“Occupational shifts” refers to net declines in employment in specific occupations Substitute teachers 154K Carpenters 40K Correctional officers and jailers 65K Maids and housekeeping cleaners 134K
between 2019 and 2022. However, we do not know exactly how individuals moved from
one occupation to another or if they made multiple moves; for that reason, we refer to the Tutors 81K Painters, construction and maintenance 25K Lifeguards, ski patrol, and 36K Coaches and scouts 26K
number of occupational shifts rather than specifying the number of workers making those other recreational protective
changes. Preschool teachers 25K Drywall and ceiling tile installers 14K Computer, automated teller, 23K
Source: O*NET; US Bureau of Labor Statistics; Current Population Survey, US Census service workers
and office machine repairers
Bureau; McKinsey Global Institute analysis
Rehabilitation counselors 25K Other categories include agriculture, creatives and art
management, mechanical installation and repair, and
McKinsey & Company property maintenance.

McKinsey Global Institute | Generative AI and the future of work in America 19

 In the wake of all this switching, more Americans are now employed in higher-wage occupations
than before the pandemic, even beyond the effects of wage inflation. Overall employment in
lower-wage occupations (which we define as those paying less than $35,000 annually) has
fallen by 3.7 percent. 34 Meanwhile, middle-wage occupations (those paying $35,000 to $57,000
annually) stayed roughly constant, with a small rise of 0.1 percent. By contrast, the number of
people in higher-wage occupations (earning more than $57,000 annually) is up 9 percent. This
slice of the labor market has added about 3.5 million jobs.
We lack clear data on how individuals moved from job to job; some may have shifted more than
once. Many Americans who changed jobs did land positions with higher pay. 35 But some portion
of the growth in higher-wage employment could be attributed to new entrants to the labor force
with higher levels of education. Some of the job losses in lower-wage occupations may be due to
people opting not to take on that work, driven by either choice or inability.
At the same time, the number of lower-wage job openings has not declined. Demand to fill
lower-wage service jobs persists, but fewer candidates want these jobs. Employers are having
hiring difficulties for many roles that do not require college degrees. Improvements in job
quality, pay, benefits, a stronger promise of career advancement, or a higher level of automation
may be necessary to resolve these shortages. As the next chapter suggests, some lower-
wage occupations (such as home healthcare aides) will be in high demand in the future, too.
Improvements in job quality, pay, benefits, a stronger promise of career advancement, or a higher
level of automation may be necessary to resolve these shortages.

Improvements in job quality, pay,

benefits, a stronger promise of
career advancement, or a higher
level of automation may be
necessary to resolve labor shortages
in lower-wage occupations.

34
We split occupations into three bands, designating the bottom 30 percent as lower wage, the middle 40 percent as middle
wage, and the top 30 percent as high wage. These designations were based on the 2022 median annual wage for each
occupation, weighted by 2022 employment. As of 2022, there were about 40 million Americans in high-wage occupations,
45 million in middle-wage occupations, and 72 million in lower-wage occupations. The prepandemic January 2020 baseline
was about 37 million, 45 million, and 75 million, respectively.
35
See, for example, “Majority of US workers changing jobs are seeing real wage gains,” Pew Research Center, July 2022.
Similarly, the Atlanta Fed’s Wage Growth Tracker shows sharply higher wage gains for workers who switch jobs than for those
who don’t; see atlantafed.org/chcs/wage-growth-tracker.aspx.

McKinsey Global Institute | Generative AI and the future of work in America 20

© Stefanamer /Getty
2. Job gains and losses
through 2030

It is impossible to know exactly what lies ahead, especially in the short term. Yet we can piece
together a picture of how multiple trends might change the mix of jobs in the US economy by the
end of the decade. 36
While automation has the biggest impact, our research incorporates a much broader range of
factors. These include shifts that were accelerated by the pandemic and appear to be lasting,
including increased remote work and virtual meetings that have reduced demand for business
travel as well as consumers embracing e-commerce. We also weigh federal investment in
infrastructure and the net-zero transition, as well as ongoing investment to digitize the economy,
rising incomes and education levels, the healthcare needs of an aging population, and the
potential marketization of unpaid domestic and care work. In addition, we assume that some
new occupations may be created at similar rates as in the past. This approach builds on previous
MGI research on the future of work as well as McKinsey’s just-published report on generative AI
more specifically. 37
The resulting analysis shows which occupations can expect to see growing demand and which
are likely to see job losses. As a result of these changes, we estimate that another 11.8 million
occupational transitions could occur by 2030. This is on top of the 8.6 million occupational
shifts that took place from 2019 to 2022. Understanding the nuances of how this might play
out and who might be affected is critical to ensuring a smooth transition for individuals and
businesses alike.

The burning question: Will generative AI be a

job destroyer for white-collar workers?
One of the burning questions of 2023 is whether generative AI might wipe out a significant share
of jobs and, if so, which ones.
Our research does not lead us to estimate job losses, although we cannot definitively rule out
that conclusion, at least in the short term. Technological advances often cause disruption—but
historically, they have eventually fueled economic and employment growth.
This research relies on aggregated employment growth projections from the US Bureau of Labor
Statistics, which assume full employment in the target year (2030). 38 While we do not predict
aggregate employment levels ourselves, we do model various drivers of labor demand to look at
how the mix of jobs might change—and those results yield some gains and some losses. In fact,
the occupational categories most exposed to generative AI could continue to add jobs through
2030, although its adoption may slow their rate of growth (Exhibit 3).

36
We do not attempt to forecast how employment may be affected by the business cycle in the short term. Our scenario offers a
longer-term structural view focused on changes in the employment mix. For a deeper discussion of the employment effects of
generative AI and federal investment in the net-zero transition and infrastructure, see chapter 3.
37
See the following McKinsey and MGI reports: The economic potential of generative AI: The next productivity frontier,
June 2023; Jobs lost, jobs gained: Workforce transitions in a time of automation, December 2017; The future of work in
America: People and places, today and tomorrow, July 2019; and The future of work after COVID-19, February 2021.
38
Projections overview and highlights, 2021–31, US Bureau of Labor Statistics, November 2022

McKinsey Global Institute | Generative AI and the future of work in America 23

Web 2023
future-of-work
Exhibit 3
Exhibit 4 and 8 of 21

While STEM, healthcare, builders, and professional fields continue to add jobs,
generative AI could change work activities significantly for many occupations.

Estimated labor demand change and generative AI Midpoint automation Employment,

automation acceleration by occupation, US, 2022–30 adoption¹ by 2030, % absolute
15– 25– 35–
35 25 35 40 5M 10M
Health
professionals

30

Health aides,
technicians,
25 and wellness
STEM
professionals

20 Increasing labor demand Increasing labor demand

and modest change of and high change of work
work activities activities

15
Builders Managers Creatives and
Change arts management
in labor
demand,² % 10 Transportation
Property
services
maintenance
Business and
legal professionals
5 Mechanical Community
Agriculture installation services Education and
and repair workforce
training
0
Production 5 10 15 20
work Food
services
–5

Decreasing labor demand

with modest change of
work activities
–10

Customer
service
–15 and sales

Office
support
–20

Increase in automation adoption driven by generative AI acceleration, percentage points

¹Midpoint automation adoption is the average of early and late automation adoption scenarios as referenced in The economic potential of generative AI: The next
productivity frontier, McKinsey & Company, June 2023.
2We consider multiple drivers affecting demand: rising income, aging populations, technology investment, infrastructure investment (including Bipartisan
Infrastructure Law), rising education levels, net-zero transitions, marketization of unpaid work, creation of new occupations, automation (including generative AI),
increased remote working and virtual meetings, and e-commerce and other virtual transactions.
Source: US Bureau of Labor Statistics; Current Population Survey, US Census Bureau; McKinsey Global Institute analysis

Even in the absence of job losses, professional roles that have largely been immune to
automation until now will still feel the effects through substantial changes in how people allocate
McKinsey & Company

their time at work. 39 In the meantime, other forces, such as the aging population and the need for

39
Generative AI is new and highly dynamic, so we acknowledge that the effects could be larger depending on adoption and the
development of new capabilities.

McKinsey Global Institute | Generative AI and the future of work in America 24

software and digital tools to continue to be deployed across the economy, will maintain demand
for healthcare, tech, and other professional roles.
It is worth noting that business and legal professions and managerial jobs grew during the
pandemic. Employers need managers to hold operations together, and it is logical that would
also be the case as new technologies redefine roles. However, within the business and legal
occupational category, some specific occupations (such as paralegals) appear likely to decline.

Net job gains for healthcare, STEM, construction, and business

and legal professionals could offset losses elsewhere
First, let’s consider the occupational categories that are resilient and growing. Healthcare
jobs are expected to have the highest demand growth, with a 30 percent increase for health
professionals and similar growth for health aides, technicians, and wellness professionals
(Exhibit 4). An aging population increases the need for specialized healthcare. In fact, four of the
ten largest job-creating occupations in our analysis are healthcare roles. Keeping the ratio of
healthcare workers to the population over age 65 roughly constant with today’s level in 2030,
we estimate that 3.8 million jobs could be added for registered nurses, personal care assistants,
home health aides, nursing assistants, licensed practical and vocational nurses, and medical
assistants. The imbalance between supply and demand is already evident. Many hospitals and
nursing homes are coping with staffing shortages that can affect the quality of care. 40
Jobs in STEM fields are estimated to grow by 23 percent by 2030. This equates to an average
annual growth rate of 2.6 percent, which represents an acceleration from the past few years. The
top-growing occupations include software developers, computer systems analysts, and data
scientists. Some of this growth is related to the deployment of automation systems themselves.
Although layoffs in the tech industry have made headlines in 2023, and generative AI can
handle some of the current work activities done by tech professionals, there is still a great need
for digital talent in sectors throughout the economy. The sheer scale of ongoing digitization,
a surge in the number of connected devices, and the growing need to guard against data
breaches should continue to fuel STEM demand. 41 In addition, the CHIPS and Science Act is
putting additional funding into R&D, scientific research, and STEM workforce development. 42 It
aims to address reliance on imported chips by directing $280 billion toward US semiconductor
capabilities, including domestic manufacturing, high-tech regional hubs, and developing the
nation’s STEM workforce. 43
The transportation services category is expected to see job growth of 9 percent by 2030,
driven by the continuing shift toward e-commerce and the delivery economy associated with
it. We also expect 12 percent growth in construction jobs as infrastructure projects break
ground nationwide.

40
Steven Ross Johnson, “Staff shortages choking U.S. health care system,” U.S. News & World Report, July 28, 2022.
41
Alan Zilberman and Lindsey Ice, “Why computer occupations are behind strong STEM employment growth in the 2019–29
decade,” US Bureau of Labor Statistics, Beyond the Numbers: Employment & Unemployment, volume 10, number 1, January
2021.
42
Note that both the CHIPS and Science Act and the Inflation Reduction Act create room for additional upside in employment.
But since uncertainty about their implementation persists as of this writing, our estimates do not explicitly incorporate their
effects on jobs.
43
“The CHIPS and Science Act: Here’s what’s in it,” McKinsey & Company, October 2022.

McKinsey Global Institute | Generative AI and the future of work in America 25

Web 2023
future-of-work
Exhibit
Exhibit 421
9 of

Healthcare, STEM, and builder roles could grow, while some occupations
requiring lower education attainment, particularly service jobs, could decline.

Estimated future US job growth and current educational attainment by occupational category
Midpoint automation scenario,¹ with generative AI acceleration

Resilient and growing occupations² Stalled but rising occupations³ Hit and declining occupations⁴

Share of workers with

Net change in labor bachelor’s degree or Employment,
Occupational category demand, 2022–30, % above,5 2022, % 2022, million

Health professionals 30 77 6.5

Health aides, technicians, and wellness 30 21 11.6

STEM professionals 23 73 7.9

Builders 12 8 7.0

Managers 11 64 9.7

Creatives and arts management 11 66 2.2

Property maintenance 10 9 4.6

Transportation services 9 11 5.6

Mechanical installation and repair 7 10 6.6

Business and legal professionals 7 68 16.0

Community services 7 44 6.8

Education and workforce training 3 75 9.9

Agriculture 2 12 2.1

Production work –1 10 13.3

Food services –2 11 13.7

Customer service and sales –13 22 14.7

Office support –18 28 20.1

¹Midpoint automation adoption is the average of early and late automation adoption scenarios as referenced in The economic potential of generative AI: The next
productivity frontier, McKinsey & Company, June 2023.
2Resilient during the pandemic, 2019–22, and expected to grow between 2022 and 2030.
3Stalled during the pandemic, 2019–22, and expected to rise between 2022 and 2030.
4Hit during the pandemic, 2019–22, and continuing to decline between 2022 and 2030.
⁵Based on 2019 demographic shares by occupation applied to 2022 employment by occupation.
Source: O*NET; US Bureau of Labor Statistics; Current Population Survey, US Census Bureau; McKinsey Global Institute analysis

McKinsey & Company

McKinsey Global Institute | Generative AI and the future of work in America 26

 In contrast to the growing categories described above, we expect the biggest declines in office
support, customer service, and food services—the same categories that took the biggest hits
during the pandemic. We estimate future losses of 1.6 million clerical jobs, for instance, plus some
830,000 retail salespeople and 630,000 cashiers (Exhibit 5). Automated systems can efficiently
handle jobs with a high share of repetitive tasks, data collection, and elementary data processing.
This trend was looming even before the arrival of generative AI.
The modest expected net losses in production work may seem puzzling in light of recent
announcements about investments in domestic manufacturing. Manufacturing growth should
boost overall US employment demand in the years ahead—but the sector is becoming more high-
tech and automated, which will change its mix of occupations. Advanced manufacturing requires
fewer jobs such as assemblers, packers, and molding machine workers but more jobs such as
software developers and industrial engineers, which fall into the STEM professional occupational
category. 44 Manufacturing also requires supporting function roles such as market research
analysts, supply chain managers, and truck drivers. 45

Even more workers may need to make occupational transitions

Compared with our previous estimates, we now see a larger number of workers potentially
needing to change occupations through 2030. A transition is needed when someone is
involuntarily displaced from a job by automation, the shift to e-commerce, the phaseout of a
high-emissions activity, or some other trend, and they are unable to get a new job in the same
occupation because demand has declined.
Some displaced workers may find alternative jobs within the same occupational category. For
example, a bookkeeping and accounting clerk could take a job as a database administrator; this
move would involve a change of occupations but keep this individual within the office support
category. But a displaced waiter who can no longer find a job in food services, a category in
decline, might need to transition to another type of work altogether. That person might become a
light truck delivery driver since demand is growing for those roles.
We find that just under 12 million additional occupational shifts could be needed by 2030
(Exhibit 6). Of those, some nine million may involve moves into entirely different occupational
categories (for example, going from a production-related job to one in transportation, or from a
customer service role to one in arts management).
People currently working in office support, customer service and sales, production, and food
service occupations—the same categories that were the source of a majority of shifts during the
pandemic—account for more than 80 percent of these potential transitions. While this could be
disruptive for many workers, it could also provide an opportunity. Many of the growing jobs that
could absorb these displaced workers are in higher wage brackets.

Just under 12 million additional

occupational transitions
may be needed by 2030.

44
For more on this topic, see Asutosh Padhi, Gaurav Batra, and Nick Santhanam, The titanium economy: How industrial
technology can create a better, faster, stronger America, Public Affairs, 2022.
45
Building a more competitive US manufacturing sector, McKinsey Global Institute, April 2021.

McKinsey Global Institute | Generative AI and the future of work in America 27

Web 2023
Exhibit 5
future-of-work
Exhibit 10 of 21
While future demand looks robust for some occupations, it is
declining for others.

Net labor demand change in top growing occupations, 2022–30, million 2022 2030
Midpoint automation scenario,¹ with generative AI acceleration

0 1 2 3 4 5 6

Personal care and home health aides

Registered nurses

Laborers and freight, stock, material movers

Management, research and operation specialists
Software developers

Nursing assistants
Truck drivers (heavy, tractor-trailer)

Construction laborers

Licensed practical and vocational nurses

General and operations managers

Net labor demand change in top declining occupations, 2022–30, million 2022 2030
Midpoint automation scenario,¹ with generative AI acceleration

0 1 2 3 4 5 6

Retail salespersons

Office clerks, general

Cashiers

Secretaries and administrative assistants²

Bookkeeping, accounting, and auditing clerks

Customer service representatives

First-line supervisors of office support workers

Receptionists and information clerks

Accountants and auditors

Shipping, receiving, and inventory clerks

¹Midpoint automation adoption is the average of early and late automation adoption scenarios as referenced in The economic potential of generative AI: The next
productivity frontier, McKinsey & Company, June 2023.
²Does not include legal, medical, and executive.
Source: O*NET; US Bureau of Labor Statistics; Current Population Survey, US Census Bureau; McKinsey Global Institute analysis

McKinsey & Company

McKinsey Global Institute | Generative AI and the future of work in America 28

Web 2023
future-of-work
Exhibit 11 of 21
Exhibit 6

Twelve million more occupational shifts could occur by 2030.

Estimated number of occupational transitions by category, 2022–30

Midpoint automation scenario,¹ with generative AI acceleration

Resilient and growing occupations² Stalled but rising occupations³ Hit and declining occupations⁴

Employment,
Occupational category Occupational transitions, absolute 2022, million
Office support 4.7M 20.1
Customer service and sales 2.7M 14.7
Production work 1.4M 13.3
Food services 1.2M 13.7
Business and legal professionals 676K 16.0
Education and workforce training 280K 9.9
Builders 243K 7.0
Mechanical installation and repair 184K 6.6
Community services 167K 6.8
Managers 130K 9.7
Agriculture 78K 2.1
Transportation services 59K 5.6
STEM professionals 30K 7.9
Creatives and arts management 27K 2.2
Health aides, technicians, and wellness 23K 11.6
Property maintenance 19K 4.6
Health professionals 15K 6.5

¹Midpoint automation adoption is the average of early and late automation adoption scenarios as referenced in The economic potential of generative AI: The
next productivity frontier, McKinsey & Company, June 2023.
2Resilient during the pandemic, 2019–22, and expected to grow between 2022 and 2030.
3Stalled during the pandemic, 2019–22, and expected to rise between 2022 and 2030.
4Hit during the pandemic, 2019–22, and continuing to decline between 2022 and 2030.
Source: O*NET; US Bureau of Labor Statistics; Current Population Survey, US Census Bureau; McKinsey Global Institute analysis

McKinsey & Company

McKinsey Global Institute | Generative AI and the future of work in America 29

© Goroden Koff / Getty
3. New forces changing
labor demand:
Generative AI and
federal investment

America is living through one of its most disruptive economic periods in decades, with pandemic-
driven changes in consumption patterns, an inflationary spike, and a shift to remote work that
changed the daily habits of a third of all workers. Against this backdrop, the arrival of ChatGPT
was another seismic event. Some users who initially tried out ChatGPT and the other generative
AI tools that quickly followed were drawn by curiosity. In subsequent months, many have already
begun to use these tools to assist in a range of work tasks.
Other forces are at work, too. One of the biggest recent developments has been the passage
of legislation that will put significant federal investment into decarbonizing the economy and
modernizing America’s infrastructure.

Generative AI and other automation technologies can

handle a growing range of work activities
Rarely has a tech innovation created such a splash as ChatGPT, which has achieved one of the
fastest adoption rates in history. 46 But what is it exactly, and how could it affect jobs? We consider
it alongside other automation technologies, many of which are already having major effects, from
robots on factory floors to automated document processing in back-office functions.
What is generative AI?
Part of the field known as deep learning, generative AI refers to applications typically built using
a class of artificial neural networks called foundation models, structures inspired by the billions of
neurons connected in the human brain. These models have taken AI applications to the next level.
ChatGPT is only one of several new generative AI tools. The “generative” in the name refers to its
ability to identify patterns across enormous sets of data and generate new content—something
that has often been considered uniquely human. ChatGPT can be used to write correspondence,
essays, articles, stories, poetry, and computer code, in addition to scraping the internet for
information and answering questions.
Another AI tool, Dall-E, can create images and illustrations in multiple styles. Dall-E and similar
image generators were “trained” by absorbing huge quantities of images and reading their
captions. They can produce new images as specified by users, in a variety of styles, and even
take stories and turn them into illustrated storyboards. Other AI systems from major platforms
can generate images, video, audio, and text.

46
Andrew R. Chow, “How ChatGPT managed to grow faster than TikTok or Instagram,” Time, February 8, 2023.

McKinsey Global Institute | Generative AI and the future of work in America 31

 Much of the buzz today is about users interacting with these systems for fun or to satisfy
their curiosity. But the potential applications for businesses are exceptionally wide-ranging.
Generative AI can be used to write code, design products, create marketing content and
strategies, streamline operations, analyze legal documents, provide customer service via
chatbots, or even accelerate scientific discovery. The ability to generate images is the first step
toward creating immersive experiences.
Generative AI is still in the early stages, and these tools are far from perfect. They can produce
inaccurate results, and many legal, ethical, and intellectual property questions still need to
be resolved.
What types of work can it do?
One of the biggest leaps forward in AI is in natural language capabilities, which are required for
a large number of work activities. Previously, these were not expected to match median human
performance for natural language understanding until around 2028 at the earliest, but they are
already reaching that point today.
These advances mean that automation is about to affect a wider set of work activities. As a
result, the share of hours spent on work tasks today across the US economy that have the
technical potential to be automated with currently demonstrated technologies has jumped from
44 percent to 62 percent in our midpoint scenario. 47
But technical potential does not always match what happens on the ground. It is the theoretical
maximum that could be automated, considering current technological capabilities at any
given time. But not everything that could be done in a theoretical world would be adopted by
businesses. The pace of actual adoption typically lags behind technical potential. It is affected
by the time needed for solution integration, whether it is economically feasible to replace human
labor with technology, and multiple other barriers such as customer acceptance, labor laws, and
companies lacking the right workforce skills.
Yet the introduction of generative AI has sharply accelerated the timeline for automation
adoption, since these tools are easy to access and integrate and are relatively low cost. Without
generative AI, we estimate that automation could take over tasks accounting for 21.5 percent of
the hours currently worked in the US economy by 2030. With it, that share jumps to 29.5 percent
(Exhibit 7). 48 This includes a significant increase in tasks involving expertise, interactions with
people, and even creativity. 49 Tasks requiring physical work, by contrast, are expected to be
less affected.

One of the biggest leaps forward

in AI is in natural language
capabilities, which are required for
a large number of work activities.

47
The midpoint scenario is the average of the early and late automation adoption scenarios referenced in McKinsey’s recent
report The economic potential of generative AI: The next productivity frontier (June 2023).
48
This is the average of a very wide range, from 3.7 to 55.3 percent.
49
Thomas H. Davenport and Nitin Mittal, “How generative AI is changing creative work,” Harvard Business Review, November
2022; and Michael Chui, Roger Roberts, and Lareina Yee, “Generative AI is here: How tools like ChatGPT could change your
business,” McKinsey & Company, December 2022.

McKinsey Global Institute | Generative AI and the future of work in America 32

Web 2023
future-of-work
Exhibit 721
Exhibit 12 of

Generative AI accelerates automation adoption in all scenarios.

Automation adoption, US, %

Early scenario with Early scenario without Late scenario with Late scenario without
generative AI acceleration generative AI acceleration generative AI acceleration generative AI acceleration

100

80

15
60 Percentage point
difference in early
scenario, 2030

40

20
1
Percentage point
difference in late
0 scenario, 2030
2021 2030 2040 2050 2060 2070 2080 2090 2100

Note: The range of scenarios reflects uncertainty regarding the availability of technical capabilities, based on interviews with experts and survey responses. The
early scenario makes more aggressive assumptions for all key model parameters (technical potential, integration timeline, economic feasibility, and regulatory
and public adoption).
Source: O*NET; US Bureau of Labor Statistics; McKinsey Global Institute analysis

McKinsey & Company

McKinsey Global Institute | Generative AI and the future of work in America 33

 Exhibit 8 shows how this might play out across occupational categories. Labor economists have
often noted that the deployment of automation technologies tends to most affect workers with
the lowest levels of skills, as measured by educational attainment; in other words, automation
has been “skill biased.” Generative AI turns that on its head, however, automating some of
the activities of knowledge workers at the higher end of the income spectrum. 50 However, it is
important to note that other types of automation technologies are continuing to affect lower-
wage occupational categories.
Computer programmers, scientific researchers, market researchers, translators, and financial
advisers are just a small sample of the specific occupations that could be affected. A recently
published report from MGI found the greatest potential economic value could be unlocked in
customer operations, marketing and sales, software engineering, and R&D. 51

50
Tyna Eloundou et al., GPTs are GPTs: An early look at the labor market impact potential of large language models, arXiv, March
2023.
51
The economic potential of generative AI: The next productivity frontier, McKinsey & Company, June 2023.
Web 2023
future-of-work
Exhibit 3 and 13 of 21
Exhibit 8
With generative AI added to the picture, 30 percent of hours worked today
could be automated by 2030.

Midpoint automation adoption¹ by 2030 as a share of time spent on work activities, US, %

Automation adoption without Automation adoption with XX — Percentage-point acceleration in

generative AI acceleration generative AI acceleration automation adoption from generative AI

0 10 20 30 40

STEM professionals 16
Education and workforce training 16
Creatives and arts management 15
Business and legal professionals 14
Managers 9
Community services 9
Office support 7
Health professionals 6
Builders 6
Property maintenance 6
Customer service and sales 6
Food services 5
Transportation services 5
Mechanical installation and repair 5
Production work 4
Health aides, technicians, and wellness 4
Agriculture 3
All sectors² 8

¹Midpoint automation adoption is the average of early and late automation adoption scenarios as referenced in The economic potential of generative AI: The next
productivity frontier, McKinsey & Company, June 2023.
²Totals are weighted by 2022 employment in each occupation.
Source: O*NET; US Bureau of Labor Statistics; McKinsey Global Institute analysis

McKinsey & Company

McKinsey Global Institute | Generative AI and the future of work in America 34
Looking in more detail at specific occupations, we see the biggest impact of generative AI
occurring for lawyers (Exhibit 9). Generative AI can search through case law for the most relevant
precedents, freeing lawyers to think through how to apply them in new legal arguments. Lawyers
also spend a great deal of time tailoring form documents, but they could use generative AI
to take the first cut and then quickly edit.52 In a completely different field, civil engineers can
use generative AI to accelerate the design process, taking all building codes into account for
fewer errors and less rework, particularly when it comes to complex mechanical, electrical, and
plumbing systems. Generative AI can also enable engineering to explore alternative designs
quickly to find the option with the best energy efficiency or the lowest use of materials. Making
engineering and architecture more productive is vital at a time when the nation needs to deliver
more affordable housing and major infrastructure projects.

52
Ilona Logvinova, “Legal innovation and generative AI: Lawyers emerging as ‘pilots,’ content creators, and legal designers,” In
the Balance blog, McKinsey & Company, May 11, 2023.

Web 2023
future-of-work
Exhibit
Exhibit 14 of921

White-collar jobs are among the most potentially impacted by generative AI.

Midpoint automation adoption by 2030 as a share of time spent on work activities, US, % by
occupation¹

Automation adoption without Automation adoption with XX — Acceleration in automation

generative AI acceleration generative AI acceleration adoption from generative AI

0 10 20 30 40 Employment,
2022, absolute

Lawyers and judges 22 824K

Math specialists 18 347K
Postsecondary teachers 18 1.6M
School teachers 18 5.6M
Building engineers 18 601K
Computer engineers 18 3.4M
Entertainers and media workers 17 1.5M
Community and social workers 15 2.5M
Business and financial specialists 15 10.4M
Engineers 14 1.6M
Account managers 14 4.2M
Architects, surveyors, and cartographers 13 199K
Gaming entertainment workers 12 119K
Scientists and academics 12 1.1M
Doctors 12 1.3M
Protective services 11 521K
Education support workers 11 1.7M
Managers 11 6.6M

1Occupations for which generative AI increased automation adoption by more than 10 percentage points in the midpoint scenario. Midpoint automation adoption
is the average of early and late automation adoption scenarios as referenced in The economic potential of generative AI: The next productivity frontier,
McKinsey & Company, June 2023.
Source: O*NET; US Bureau of Labor Statistics; McKinsey Global Institute analysis

McKinsey & Company

McKinsey Global Institute | Generative AI and the future of work in America 35
Changing the way people work
Before the arrival of generative AI, the pandemic accelerated the adoption of automation systems
as many employers scrambled to respond. For greater flexibility to meet erratic demand, for
example, retailers including Amazon, Walmart, and Target started to enlist industrial robots to
pick, sort, and track merchandise in their warehouses. Restaurants shifted to automated order
taking, while many retailers opted for self-checkout and touchless technology.
It is important to note that automation adoption is not the same as eliminating jobs. Many jobs
with some automatable tasks will remain, but the day-to-day nature of what people do and how
they do it changes. In fact, employment demand should continue to rise in many occupations
most exposed to generative AI, although perhaps at a slower rate.
Generative AI and other automation technologies could help the United States reignite its
flagging productivity growth, partially compensating for declining employment growth as
the population ages. 53 Capturing that effect means that workers would have to use these
technologies properly and effectively.
Importantly, individuals would need to use the time that is freed up to focus on higher-value
activities. Managers, for instance, could rely on automation for mundane administrative and
reporting tasks and use the time saved to provide more one-on-one coaching. Researchers
could dramatically reduce the time they spend sorting and synthesizing data or conducting
literature searches of existing studies, spending more time on original contributions and
speeding research projects. Teachers could use generative AI to grade tests and flesh out lesson
plans (at least partially) and focus more of their energy on interacting with students.54
Things may be produced faster, potentially with smaller teams. If generative AI can create
intricate storyboards, animated films could be produced with fewer artists, although artists would
still provide the ideas and shape what the tools produce. Generative AI can compose technical
writing, but writers will still need to report, edit, think, and apply real-world judgment. At the
moment, these tools might be best used to produce first drafts in creative fields—but assistance
with first drafts can be a major productivity lift.
While automation tends to dominate the conversation when it comes to the future of work, it is
not the only force at work. The remainder of this chapter will touch on the ways that investment
priorities shape the labor market.

An infusion of federal investment is poised to shift employment

demand to greener sectors and boost construction
While technology is evolving, other priorities are coming to the forefront. The United States has
recently passed major funding bills that will change the scope of employment demand across the
country, including the Inflation Reduction Act, the Bipartisan Infrastructure Law, and the CHIPS
and Science Act.
Funding for the net-zero transition
The United States, along with other countries, has made a formal international commitment
to address climate change by reducing greenhouse-gas emissions to net zero by 2050 at the
latest. 55 Meeting this target requires extensive groundwork to transform everything from energy
generation and transmission to transportation, industry, and agriculture. These changes are
already getting under way and will continue to do so through 2030. The Inflation Reduction
Act, passed in August 2022, adds both public funding and momentum, although its precise
implementation was not clear at the time of our analysis.

53
Rekindling US productivity for a new era, McKinsey Global Institute, February 2023.
54
Jake Bryant, Christine Heitz, Saurabh Sanghvi, and Dilip Wagle, “How artificial intelligence will impact K-12 teachers,”
McKinsey & Company, January 2020.
55
See “Raising ambition: Net-zero coalition,” United Nations, accessed June 25, 2023.

McKinsey Global Institute | Generative AI and the future of work in America 36

 The net-zero transition could reallocate a substantial number of American jobs. To estimate
these effects, we build on previous McKinsey research assessing the impact at a global level and
zoom in on the United States. 56 We use a scenario-based analysis, drawing on the approach used
by the Network for Greening the Financial System. 57
We estimate that the effort to get to net zero could create a net gain of some 700,000 jobs in
the United States by 2030. But the churn could be far greater than this number implies. Some
3.5 million jobs could be lost through direct and indirect effects across the economy. This should
be more than offset by a gain of 4.2 million positions, primarily led by capital expenditures
on renewable energy. 58 However, the new jobs may be in different locations and require
different skills.
Our estimates point to job growth in fields such as renewable power generation and storage—
positions such as wind turbine technicians, electrical engineers, and solar photovoltaic installers.
Capital spending to build low-emissions facilities and retrofit existing plants and buildings
with cleaner technologies is expected to drive demand for construction and specialized
manufacturing workers. At the same time, the economy will need fewer jobs in carbon-intensive
industries, particularly oil, gas, and mining. Some of these shifts are already playing out: the US
solar industry currently employs twice as many workers as the nation’s coal industry, and it is
struggling to hire quickly enough to keep up with its rapid expansion.59
Our findings are based on an “orderly transition” scenario—and it is important to note that the
impact on the labor market could be more severe if the transition is “disorderly” (for example,
if the shift away from fossil fuels occurs before a sufficient and reliable supply of renewables
is ready to deliver). Power disruptions and spikes in energy prices would have economic ripple
effects that could lead to more abrupt job losses.
To ensure that the availability of workers does not become a bottleneck to achieving its climate
goals, the United States will need to develop the skills required for the jobs that will be in demand
in a greener economy. Training and redeployment will be especially important in regions where
industries such as oil refining and traditional auto manufacturing are major employers. Specific
communities could face significant displacement as legacy industries shrink, while others that
seed green industries could enjoy more robust job growth. Geographic mismatches and the
availability of retraining programs will determine to a large extent whether this process is smooth.
Identifying transferable skills can illuminate some of the most viable pathways and inform
retraining efforts (Exhibit 10).

The net-zero transition

could reallocate a substantial
number of American jobs.

56
This research analyzes the magnitude of what is needed in 69 countries that account for about 95 percent of global GDP. The
net-zero transition: What it would cost, what it could bring, McKinsey Global Institute, January 2022.
57
NGFS Net Zero 2050 scenario using REMIND-MAgPIE (phase 2). Under the scenario, the United States reaches net-zero
emissions across all greenhouse gases.
58
This analysis assumes that the supply chain composition remains the same by 2030. We consider the impact of the United
States meeting its own emission targets as well as its role in meeting global demand for products. We consider both direct
and indirect jobs.
59
National solar jobs census 2021, Interstate Renewable Energy Council, July 2022.

McKinsey Global Institute | Generative AI and the future of work in America 37

Web 2023
future-of-work
Exhibit
Exhibit 15 10
of 21

Transferable skills can help workers move into growing and higher-paying jobs.

FROM Transferable skills New skills added TO

Automotive Wind turbine
technician technician

$46,970 $57,320
annual salary annual salary
• Complex problem solving • Use of hydraulic, power
Postsecondary • Customer service torque, and diagnostic Postsecondary
• Troubleshooting tools certification
certification
• Field service management • Working knowledge of
664K jobs industrial safety Inspect,
in 2022 practices diagnose,
• Ability to work with adjust, or repair
diverse team wind turbines
• Ability to use technology
apps and devices

FROM Transferable skills New skills added TO

Building maintenance HVAC contractor,
technician heat pump expertise

$44,980 $51,390
annual salary annual salary
• Ability to diagnose • Design installation,
High school and repair HVAC and maintenance and repair Specialized
electrical systems of residential heat pump training
diploma
• Strong mechanical/ systems
1.5M jobs technical aptitude • Proficiency with Ensure that
in 2022 • Understanding of construction management HVAC/R
blueprints and schematics software systems meet
• Knowledge of diagnostic • Knowledge of energy specifications
tools such as multimeters efficiency codes for energy efficiency

Source: EMSI Burning Glass; US Bureau of Labor Statistics; Lightcast; Indeed.com; IREC (2022) Green Buildings Career Map; McKinsey Global Institute analysis

McKinsey & Company

Renewing America’s infrastructure

Recent federal commitments to expand and revitalize infrastructure could boost employment
and eventually improve the overall productivity of the US economy. But labor shortages in the
construction and manufacturing sectors will need to be addressed in order to achieve these
national priorities.60
The Bipartisan Infrastructure Law, signed in 2021, includes $1.2 trillion in federal funding, almost
half of which is new spending. This investment will go to a variety of projects nationwide: repairing
roads and bridges, rebuilding water infrastructure to eliminate lead pipes, expanding high-
speed internet, improving public transit and passenger rail, modernizing ports and airports, and
building a national electric vehicle charging network.61 To put this into perspective, the legislation
will boost federal infrastructure spending during the next five years to almost the same average
share of GDP spent during the New Deal from 1933 to 1937.62

60
Garo Hovnanian, Adi Kumar, and Ryan Luby, “Will a labor crunch derail plans to upgrade US infrastructure?” McKinsey &
Company, October 2022.
61
“President Biden’s Bipartisan Infrastructure Law,” The White House, November 2021.
62
Adie Tomer, “The Senate infrastructure bill puts America close to another New Deal,” Brookings Institution, August 2021.

McKinsey Global Institute | Generative AI and the future of work in America 38

Box 1
Seizing the electric vehicle opportunity

Transportation accounted for a larger share (37 percent) of need to repurpose engineering talent toward electrical and
the nation’s CO2 emissions in 2021 than any other sector.1 electronics specialties.
Switching to electric vehicles (EVs) is therefore critical to
EVs run on thousands of chips—in fact, roughly double the
getting to net zero. The federal government has set a goal
number found in conventional vehicles. 4 Manufacturers ran
that half of all vehicles sold by 2030 should be electric; it
into shortages in 2022, and McKinsey has estimated that
recently committed $7.5 billion to help set up a national
automotive semiconductor demand could rise by about
charging system as part of the Bipartisan Infrastructure Law
11 percent annually through 2030. The CHIPS and Science
and established tax credits for the purchase of EVs as part
Act of 2022 aims to address US reliance on imported
of the Inflation Reduction Act.2
chips by directing $280 billion toward US semiconductor
This will be a massive shift, creating disruption for current capabilities, including domestic manufacturing, high-tech
market participants and legacy automotive employees. EVs regional hubs, and developing the nation’s STEM workforce.
have fewer moving parts than traditional internal combustion Intel has broken ground on a new $20 billion semiconductor
engines, which makes their production less labor-intensive. factory in Ohio; this is on top of $20 billion to expand
This may result in job losses for assembly line workers, its production footprint in Arizona, a state that has also
molders, and machinists. captured one of the largest foreign investments in US history
from chip giant TSMC. 5
Even as demand falls for traditional automotive production
roles, automakers will need more tech-oriented workers, New types of jobs associated with EVs are already growing
from software developers and IT managers to computer as production takes root in states including California,
hardware engineers. EVs require expertise in chemical Georgia, Michigan, North Carolina, and Tennessee.
and electronic engineering in their design, testing, and Established automakers are opening battery hubs
repair. 3 The industry will have a heavier emphasis on in Kentucky, Michigan, Nevada, North Carolina, and
electronics, mechatronics and calibration, advanced Tennessee. Additional jobs will be created to produce and
materials, and additive manufacturing. There may be a install charging equipment and infrastructure nationwide.

1
“Energy and the environment explained: Where greenhouse gases come from,” US Energy Information Administration, 2022.
2
“Fact sheet: Biden-Harris administration announces new standards and major progress for a made-in-America national network of electric vehicle chargers,” The
White House, February 15, 2023.
3
Manuel Altmeier, Harald Bauer, Anna-Christina Fredershausen, and Markus Simon, Electromobility’s impact on powertrain machinery, McKinsey & Company, May
2021.
4
David Ferris, “Chip shortage threatens Biden’s electric vehicle plans, Commerce secretary says,” Scientific American, November 2021.
5
“Intel announces next US site with landmark investment in Ohio,” Intel, January 21, 2022; “Intel breaks ground on two new leading-edge chip factories in Arizona,”
Intel, September 24, 2021; and Cecilia Kang, “How Arizona is positioning itself for $52 billion to the chips industry,” New York Times, February 22, 2023.

In 2022, the infrastructure law was followed by the Inflation Reduction Act, which includes an
additional $369 billion in complementary funding. Its provisions are meant to spur domestic
manufacturing and energy efficiency upgrades, decarbonize and shore up the US electric grid,
establish a national green bank, and improve resilience in agriculture and rural areas.63 However,
both pieces of legislation establish myriad programs, with much of the funding directed toward
state and local governments through competitive grants. Coordination and implementation will
be complex, and the full impact will take time to manifest.64
From a labor market perspective, the United States will need to address its existing shortage of
construction workers to realize these plans. Employment has stalled, but demand is growing. We
estimate that jobs in the construction sector could increase by 12 percent by 2030, translating to

63
Savannah Bertrand, “How the Inflation Reduction Act and Bipartisan Infrastructure Law work together to advance climate
action,” Environmental and Energy Study Institute, September 2022.
64
See, for example, “Delivering on the promise of federal infrastructure funds in states,” McKinsey & Company, May 2022; and
“A new era of infrastructure grants,” McKinsey & Company, May 2022.

McKinsey Global Institute | Generative AI and the future of work in America 39

 more than a million net jobs gained, in addition to growth in affiliated sectors. The construction
sector could also see significant increases in labor demand from other factors, such as the need
to rebuild in the aftermath of natural disasters.65
In the peak years of infrastructure spending, about half of the demand for labor may come from
contractor roles in the construction industry itself. Another large share, just over 40 percent,
is likely to come from the materials sector, which combines manufacturing, distribution, and
warehousing.66 This part of the value chain is equally important, since delays and shortages in
obtaining materials can grind construction projects to a halt or drive up their costs.
In terms of specific occupations within the sector, our estimates indicate that construction
laborers (with 240,000 positions added), supervisors (150,000), and electricians (130,000) could
experience the largest growth in demand by 2030.
Yet the United States already has a shortage of construction workers. Some 490,000 positions
were unfilled in December 2022—the highest number recorded since industry-level data
collection began in 2000. This number dropped slightly to roughly 383,000 positions in May
2023. In September and December 2022, unfilled construction positions exceeded the number
of unemployed individuals seeking construction jobs. This imbalance is caused by a mix of
short-term, cyclical, and more structural factors. The workforce is getting older, with the share
of construction workers age 55 and older having doubled, from 11 to 22.7 percent, between
2003 and 2022. The industry is also heavily reliant on immigrants (who represent more than
a quarter of construction workers) at a time when immigration is only now picking up after a
drop of several years. In addition, women make up only about 10 percent of the construction
sector’s workforce.67
Most infrastructure development companies are competing with other industries when it comes
to skilled trades. A large number of welders, for example, will be needed in both manufacturing
and construction, and without cooperation to train new talent, competition for this small pool
of skilled tradespeople will only increase. In fact, people with skills crucial to the construction
industry have broader options. There is competing demand in both construction and logistics for
truck drivers and freight movers, who are already in short supply. The same goes for electricians,
who can work for construction companies or utilities.
Addressing these labor shortages is an opportunity to diversify the construction workforce.
According to the BLS, nine out of ten US construction workers are men; 30 percent are Hispanic;
and almost 61 percent are non-Hispanic White. Recruiting and training more women and people
of color can help companies meet the coming demand while diversifying their own workforces
and expanding opportunity more broadly.

Most infrastructure development

companies are competing
with other industries when
it comes to skilled trades.
65
See, for example, Maria Sacchetti, “Florida needs workers to rebuild after Ian. Undocumented migrants are stepping in,”
Washington Post, October 23, 2022; and Tatiana Flowers, “Supply and labor shortages double the time it’ll take to rebuild
homes lost in the Marshall fire,” Colorado Sun, January 13, 2022.
66
Garo Hovnanian, Adi Kumar, and Ryan Luby, “Will a labor crunch derail plans to upgrade US infrastructure?” McKinsey &
Company, October 2022.
67
“The construction industry: Characteristics of the employed, 2003–20,” US Bureau of Labor Statistics, Spotlight on
Statistics, April 2022.

McKinsey Global Institute | Generative AI and the future of work in America 40

© Maskot/Getty
4. Who’s vulnerable?

At the heart of it, displacement and change affect workers and their families in significant ways.
Some of the people most likely to be affected by automation are already living paycheck to
paycheck, and for them, even a short period of disruption could provoke tremendous stress.
Millions of Americans may need to switch occupations to find new ways to support themselves.
Designing interventions starts with identifying exactly who is at risk of displacement. Overlaying
our analysis showing which occupations have the most automatable activities with data on
who holds those jobs today paints a picture of uneven effects, since many occupations have
skewed demographics.
But what lies ahead is not only about risk; it is also about opportunity. With low-wage jobs
decreasing, more people could land better-paying jobs if they can access skills training and if
employers are willing to take more chances in hiring. The United States has no time to lose in
setting up large-scale retraining programs and better job-matching systems to help people make
successful transitions.

The decline of low- and middle-wage jobs could

continue as more high-wage jobs are added
The loss of middle-wage jobs and the resulting polarization in the job market is a well-
documented historical trend.68 While the share of middle-wage jobs is likely to continue declining,
our analysis finds that lower-wage jobs could see an even faster decrease, especially as office
support and customer service and sales jobs continue to disappear (Exhibit 11).

With low-wage jobs decreasing,

more people could land
better-paying jobs if they can
access skills training and if
employers are willing to take
more chances in hiring.

68
See, for example, David H. Autor, The polarization of job opportunities in the U.S. labor market: Implications for employment
and earnings, Center for American Progress and The Hamilton Project, Brookings Institution, April 2010.

McKinsey Global Institute | Generative AI and the future of work in America 43

Web 2023
future-of-work
Exhibit
Exhibit 16 11
of 21

As the US economy adds more high-wage jobs, the shares of low and
middle-wage jobs could decline.

Change in US employment by wage band,1, 2 2022–30, jobs share change, percentage points

High wage
2.6
Annual salary: >$57K

Middle wage
$35K–$57K –0.3

Low wage
–2.3
<$35K

Employment distribution of occupational categories by wage terciles,1, 2 2022, %

High wage Middle wage Low wage

Managers 97 3
Health professionals 91 6 3

STEM professionals 86 14 0.6

Business and legal professionals 79 21

Creatives and arts management 48 42 10

Education and workforce training 20 55 26

Builders 12 81 7

Community services 6 63 31

Health aides, technicians, and wellness 3 23 73

Office support 3 33 64

Mechanical installation and repair 2 90 7

Customer service and sales 2 11 87

Transportation services 2 43 56

Production work 0.8 30 70

Agriculture 12 88

Food services 1 99

Property maintenance 8 92

Note: Figures may not sum to 100%, due to rounding.

1Based on 2022 real wages (base year = 2010).
2Lower-wage jobs are those below the 30th percentile; mid-wage jobs are those between the 30th and 70th percentiles; and high-wage jobs are those above
the 70th percentile of wage distribution. Growth from 2022 to 2030 holds 2022 wage categorization constant.
Source: O*NET; US Bureau of Labor Statistics; Current Population Survey, US Census Bureau; McKinsey Global Institute analysis

McKinsey & Company

McKinsey Global Institute | Generative AI and the future of work in America 44

 While the individuals in low-wage jobs today are most vulnerable due to shrinking demand in
office support and customer service and sales jobs, this also presents an opportunity to help
them land better-paying jobs. Looking at this shift from a different vantage point, this could
be a positive opportunity for more inclusive growth as the US economy rebalances toward
higher-wage jobs (Exhibit 12). As we will discuss further in chapter 5, training and less rigid hiring
approaches will be essential to creating those pathways.
Despite this overall rebalancing toward higher-wage jobs, it is important to note that some of the
fastest-growing occupations are still some of lowest paying in the economy. Several of these are
roles that provide vital social and economic infrastructure—most notably home healthcare aides,
nursing assistants, and childcare workers. These jobs are disproportionately held by women
(Exhibit 13). Meeting these needs, especially at a time when workers have more options, may
depend on upgrading the quality of those jobs. Indeed, a shortage of childcare and elder care is
already keeping some people out of the workforce.
Web 2023
future-of-work
Exhibit 17 of 21
Exhibit 12

The economy could reweight toward higher-wage jobs.

Potential impact of occupational transitions on wage distribution,1 million

Highest wage 1st

0.2 0.2
3.0
1.3
2nd 1.3
0.8

0.8 2.5
3rd 0.6

Wage
quintiles 2.5 1.5

4th 4.3 1.1

0.4 1.1

1.1

Lowest wage 5th 5.2

3.7 3.7

2022 2030

Note: Figures may not sum to 100%, due to rounding.

1We expect 11.8 million occupational transitions by 2030. We refer to people needing to make transitions if demand is projected to decline in their current
occupation. Assumes workers move to jobs in higher-wage quintiles if there are not enough jobs in their current wage quintile.
Source: O*NET; US Bureau of Labor Statistics; Current Population Survey, US Census Bureau; McKinsey Global Institute analysis

McKinsey & Company

McKinsey Global Institute | Generative AI and the future of work in America 45

Web 2023
future-of-work
Exhibit 18 of 21
Exhibit 13

Women are underrepresented in high-growth and high-wage jobs.

Female representation in occupations projected for greatest job growth, 2022–30, US

Midpoint automation scenario,¹ with generative AI acceleration

High female representation2 Low female representation2

2,000

Below Above
average average
wage wage

Health aides
and health
support

Nurses, physicians
assistants, and
1,500
pharmacists
Personal
care workers

Net job
growth
2022–30,
thousand

Computer engineers
1,000
Material
movers and Managers
loaders

Construction
workers

Community and social workers

Building and Health technicians
grounds Installation and repair workers
cleaners (buildings and infrastructure)
500
Computer support workers
Scientists and academics
Fine equipment Therapists
installation and
repair workers Engineers
Building engineers
Childcare Doctors
Computer specialists
workers
Math specialists
Personal Artists and
appearance designers
workers Architects, surveyors, and cartographers
0

0 50 100 150 200

Wages 2022,3 $ thousand

¹Midpoint automation adoption is the average of early and late automation adoption scenarios as referenced in The economic potential of generative AI:
The next productivity frontier, McKinsey & Company, June 2023.
2High representation = equal to or above 50% in 2022; low = below 50%.
3Based on 2022 real wages, base year = 2010.
Source: O*NET; US Bureau of Labor Statistics; Current Population Survey, US Census Bureau; McKinsey Global Institute analysis

McKinsey & Company

McKinsey Global Institute | Generative AI and the future of work in America 46

Workers in low-wage jobs and those without college degrees
are more likely to need to change occupations by 2030
Based on the demographics of various jobs today, workers in lower-wage roles, those with less
education, women, and people of color are disproportionately represented among those who
may need to find new types of work by 2030 (Exhibit 14). Workers with these profiles are more
vulnerable to job displacement and will need support to make successful job transitions.
— Workers in lower-wage jobs are up to 14 times more likely to face the need to change
occupations. We estimate that almost 80 percent of all potential transitions (affecting some
9.4 million people) could occur among workers in the two lowest quintiles of wages—that
is, people earning less than $38,200. Retail salespeople, cashiers, and stockers and order
fillers all fall within the lowest quintile and could account for some 18 percent of all potential
transitions by 2030. But notably, the most vulnerable group overall is not the very lowest
wage bracket; it is those earning between $30,800 and $38,200. This group specifically is
14 times more likely to need to find new roles by 2030 than the highest-wage earners.
— Workers without college degrees are 1.7 times more likely to face displacement and
the need to change occupations than those with bachelor’s degrees or above. It is well
established that workers with a high school diploma or less have experienced limited job
prospects in the United States.69 Assuming the distribution of educational attainment levels
across occupations remains the same as in 2022, workers without bachelor’s degrees could
account for about 8.9 million of the estimated 11.8 million occupational shifts (roughly three-
quarters of the total). Notably, 90 percent of production workers are not college educated,
and about 1.4 million may need to change fields by 2030. Similarly, 72 percent of office
support workers lack college degrees, and about 4.7 million may need to move to different
occupations. Higher levels of educational attainment remain strongly positively correlated
with being in less automatable roles.
— The youngest and oldest workers are 1.4 and 1.1 times more likely to face occupational
changes, respectively. In our estimates, workers between the ages of 18 and 24 and those
above age 50 are more likely to face job transitions than workers ages 25 to 49. As younger
workers are just starting their careers and many are likely still attending school, they may be
better positioned to adapt and find different pathways. But older workers may find it harder to
start new chapters.
— Women are 1.5 times more likely to face occupational transitions than men. Women could
face 57 percent of all the transitions envisioned in our estimates, which is greater than their
overall workforce representation of 47 percent. This reflects the fact that women are heavily
represented in office support and customer service roles, which could shrink by about
3.7 million and 2.0 million jobs, respectively, by 2030.
— Hispanic workers are 1.1 times more likely to need to change occupations than White
workers. Based on the jobs they hold today, Hispanic Americans could account for 2.1 million
potential job transitions (18 percent of the total through 2030). Much of this is driven by
relatively higher representation in automatable food service roles.
— Black workers are 1.1 times more likely to face occupational changes than White workers.
Some 1.5 million Black Americans could face the need to change roles, accounting for
12 percent of all estimated transitions. While the magnitude may seem relatively small, Black
Americans have experienced higher unemployment rates than White workers since the
Census Bureau Current Population Survey began publishing data in 1972. Ensuring that Black
workers in shrinking occupations have pathways to move into high-quality jobs can help to
relieve labor supply constraints while creating a more inclusive economy.

69
See, for example, Ariel J. Binder and John Bound, The declining labor market prospects of less-educated men, NBER
working paper number 25577, February 2019.

McKinsey Global Institute | Generative AI and the future of work in America 47

Web 2023
future-of-work
Exhibit 19 of 21
Exhibit 14

Workers in lower-wage jobs, those who have less education, and women are
more likely to face displacement and need to change occupations.

Estimated occupational transitions by 2030, % transition of employment within each category1

Midpoint automation scenario, with generative AI acceleration
Race/
Wage2 Education Age Gender ethnicity

15

Low

More
likely to 14.3×
change likelihood for
jobs low-wage
workers to be
12 affected than
highest-wage
workers

1.4×
more likely
1.5×
Lowest 1.7× more likely
18–24
more likely 1.1×
9
Female more likely
Below Black
bachelor’s 50+
National degree Hispanic
average:
7.5%
White

25–49
6 Asian
Male

High Bachelor’s
degree
or above
Middle

3
Less
likely to
change
jobs

Highest

1Average national transition rate = 7.5%, calculated as total estimated transitions by 2030 divided by total employment in 2022.
2We split occupations into quintiles according to their wage distributions. Highest ≥ $68,700; high = $49,500–$68,700; middle = $38,200–$49,500; low
$30,800–$38,200; lowest < $30,800. These designations are based on 2022 average annual real wages (base year = 2010) associated with each occupation,
weighted by 2022 employment.
Source: O*NET; US Bureau of Labor Statistics; Current Population Survey, US Census Bureau; McKinsey Global Institute analysis

McKinsey & Company

McKinsey Global Institute | Generative AI and the future of work in America 48

 The workforce of the future will need more technological and
people skills—but physical work is not disappearing
To assess the potential shift in demand for workforce skills, we look into the change in skill mix
for each occupation as well as the change in demand for each occupation, and then aggregate
them for the country. To assess the skill mix of each occupation, MGI used about 2,100 individual
work activities that make up more than 850 occupations from O*NET. We mapped each activity
to a specific skill within five broad categories: physical and manual, basic cognitive, advanced
cognitive, social and emotional, and technological. The results reaffirm the trends identified in
our earlier research.
Social and emotional skills and technological skills
Given the impact of automation on individual activities as well as the shifting mix of occupations
projected in our estimates, we find that social and emotional skills and technological skills will see
the biggest increase in demand by 2030. These types of skills are particularly important in the
healthcare and STEM occupations that are poised for strong growth. However, it is possible that
generative AI could change our definitions and the way these skills are deployed over time.
Social and emotional skills (such as empathy) have always been critical in healthcare, and that
will remain paramount. But now healthcare delivery increasingly involves the use of digital
systems as well. One study found that digital content in healthcare roles like registered nurses
rose by 40 percent or more between 2002 and 2016.70 That is in line with the generally observed
increase in digital content across jobs in that period.
Skill shifts within occupations present some opportunities for people to direct more of their
time to higher-value work. For example, retail employees could spend less time processing
payments and more time assisting customers and delivering a superior experience. Healthcare
workers could reallocate time spent filling out insurance paperwork to interacting with patients.
Importantly, managers can save time on administrative work, freeing them up for more coaching
and collaboration. Any potential increase in managerial attention to team building and developing
people is important, especially at a time when being valued by a boss and having a sense of
belonging are important factors for employees weighing whether to stay in their current jobs.71

Skill shifts within occupations

present some opportunities for
people to direct more of their
time to higher-value work.

70
Mark Muro et al., Digitalization and the American workforce, Brookings Institution, November 2017.
71
Aaron De Smet, Bonnie Dowling, Marino Mugayar-Baldocchi, and Bill Schaninger, “‘Great Attrition’ or ‘Great Attraction’? The
choice is yours,” McKinsey Quarterly, September 2021.

McKinsey Global Institute | Generative AI and the future of work in America 49

 Basic cognitive or physical skills
By contrast, demand will shrink for activities that primarily require basic cognitive skills
(Exhibit 15). More of the routine tasks associated with office support roles can be handled by
software and AI, for example, while robots are able to handle tasks in warehouses and assembly
lines with greater dexterity and accuracy. People who need to move out of declining lower-wage
occupations that emphasize manual and basic cognitive tasks could have more choices—
including better-paying options—if they can interact with people and digital systems.
The share of physical work in the US economy could decline slightly, accounting for about
31 percent of total skills deployed by 2030.
The recent increase in transportation services employment seems likely to endure, and a
renewed national focus on infrastructure, climate, and resilience is set to add construction and
manufacturing jobs, as described in the previous chapter. Some healthcare roles similarly require
physical skills. The sustained importance of physical and manual skills is also informed by the fact
that robots do not yet have the sensory abilities and flexibility of the human hand; people with fine
motor skills can still do many things that machines cannot.
Additionally, those who stay in the same line of work or even in the same job are not immune to
change. Technology will alter the day-to-day tasks most Americans perform in the course of
their jobs. Staying relevant in this rapidly changing environment will mean that workers need to
continually learn new skills.
Web 2023
future-of-work
Exhibit
Exhibit 2015
of 21

To move into higher-wage occupations, workers will need to be trained more on

social and emotional skills and technological skills.

Time spent using various types of skills by wage quintile in the United States,¹ 2030, %
Midpoint automation scenario, with generative AI acceleration

Basic Higher Physical and Social and Technological

cognitive cognitive manual emotional skills
skills skills skills skills
Highest
wage quintile 1st 7 24 3 40 26

2nd 9 33 10 31 17

3rd 12 21 40 14 14

4th 21 18 35 13 13

Lowest wage
5th 17 12 52 12 8
quintile

Note: Figures may not sum to 100%, due to rounding.

1We analyzed the skills needed across a workforce by reviewing around 850 occupations and 2,100 activities, and then classifying each work activity according
to the primary skill used among a set of 25 skills, which are further categorized into 5 skill categories mentioned. We use the methodology in this report that
was first developed in MGI’s research in Skill shift: Automation and the future of the workforce, May 2018.
Source: O*NET; US Bureau of Labor Statistics; Current Population Survey, US Census Bureau; McKinsey Global Institute analysis

McKinsey & Company

McKinsey Global Institute | Generative AI and the future of work in America 50

© Lamontak590623/Getty
5. Preparing for the
future of work

As automation and other structural trends continue to raise demand for some occupations and
lower it for others, more turbulence could lie ahead. The Great Attrition caught many employers
by surprise. But mobility can be a healthy dynamic if it’s fueled by growth and leads people to
better matches where they can realize their potential.
Leaders in the business, policy, and education domains can prepare for what comes next by
keeping employees engaged, helping them add more advanced skills, making training widely
available, and bringing more people off the sidelines and into the workforce. Instead of watching
people drop out of the workforce from discouragement, stakeholders will need to help midcareer
workers reinvent themselves and pull in more of the people who are currently on the sidelines.
The nation needs clear pathways for people to enter roles that are critical to its priorities, such as
healthcare and skilled trades involved in infrastructure and advanced manufacturing.
While many new initiatives have gotten under way to help workers add new skills and change
career paths, this is a moment to double down on what works and commit to creating a more
inclusive future with greater opportunities for a bigger slice of the workforce.

The workplace plays a powerful role in

developing the nation’s human capital
While much of the debate over the future of work focuses on the need for large-scale training
programs, a great deal of skills development happens day-to-day on the job. Previous MGI
research on human capital examined the job moves made by millions of workers over a decade
and found that work experience contributes 40 percent of the average individual’s lifetime
earnings in the United States. Skills learned on the job are an even bigger determinant for people
with lower educational credentials who start out in lower-wage work.72
MGI’s research showed that almost one-third of US workers are on a path to move into higher
earning brackets, and amassing work experience in an effective way seems to be their key
to upward mobility. Each job move typically increased wages—and the bolder the move, the
bigger the boost. Companies with the strongest organizational health, those that offer more
coaching and training for their employees, and those that provide more opportunities for internal
advancement seem to help increase lifetime earnings.73
Most employers can benefit from expanding the lens through which they evaluate candidates.
Instead of insisting on prior experience that best matches the responsibilities in an open role,
organizations can shift toward evaluating candidates on their capacity to learn, their intrinsic
capabilities, and their transferable skills. This requires removing biases that pigeonhole people
into the roles they are already performing. In our sample, more than half of all role moves
undertaken by individuals involved a substantial share of new skills, which implies that most
people are capable of learning and doing more than they are doing today. Employers can benefit
from being more open to hiring candidates who have taken unconventional career paths.

72
Human capital at work: The value of experience, McKinsey Global Institute, June 2022.
73
Ibid.

McKinsey Global Institute | Generative AI and the future of work in America 53

 Capturing the productivity potential of automation, including
generative AI, requires changing the way people work
The United States urgently needs to jump-start its flagging productivity growth, which has been
a lackluster 1.4 percent since 2005. Recent MGI research noted that the economy faces looming
headwinds, including workforce shortages, debt, inflation, and the cost of the energy transition—
and higher productivity will make it easier to navigate these challenges.74
Technology is one of the critical levers, and generative AI in particular is opening up many
possibilities for making knowledge work more productive. But employers need to have a
sustained strategy for implementation that includes complementary investment in talent and
ecosystems, plus a willingness to update organizational models. Studies have shown that this
process can create a productivity “J-curve” over time, with small initial benefits eventually
yielding much bigger long-term value for businesses that stay the course.75
To capture the full benefits, employers and policy makers would need to establish clear
guidelines and guardrails for the responsible use of generative AI. Businesses will need to think
through how jobs and responsibilities could evolve, and workers would need to see these tools
not as job destroyers but as work enhancers. When machines take over dull or unpleasant tasks,
people can be left with more interesting work that requires creativity, problem solving, and
collaborating with others (Exhibit 16).
Importantly, workers would need to become proficient in using AI—and use the time that is freed
up to focus on higher-value activities. Leaders can spend more time on strategic thinking and
coaching when they automate more of their administrative and reporting tasks. Researchers
could speed up projects by relying on automation tools to sort and synthesize large data sets.
Writers can use generative AI to produce first drafts, then spend time refining and shaping
the output.

When machines take over

dull or unpleasant tasks,
people can be left with more
interesting work that requires
creativity, problem solving, and
collaborating with others.

74
Rekindling US productivity for a new era, McKinsey Global Institute, February 2023.
75
Erik Brynjolfsson, Daniel Rock, and Chad Syverson, The productivity J-curve: How intangibles complement general-purpose
technologies, NBER working paper number 25148, October 2018.

McKinsey Global Institute | Generative AI and the future of work in America 54

Web 2023
future-of-work
Exhibit 16
Exhibit 21 of 21

The evolution of a software developer’s job shows how generative AI can boost
productivity.

Illustrative

Today, Serena works as a Current mix of activities

software developer Collaborating and Writing
She shifts among 21 unique activities. communicating with software
technical colleagues code

Serena’s company adopts

new technologies
The company sets guidelines for using R&D and
generative AI to write code. innovation
The company also invests in cloud
Code
infrastructure that enables easier
Assessing correction
collaboration across geographies as well
as a version-control system that database
streamlines team projects. performance

Future mix of activities

By 2030, Serena is more productive Some tasks expand, such Some tasks are
The automation of some of Serena’s as communicating and automated, such
activities allows her to focus on more collaborating with as writing initial
valuable tasks and brings new colleagues across the drafts of code
opportunities to her firm. The activities organization
that make up her work shift.

R&D and
innovation

And her firm is larger and

more versatile Code
Technological advances enable the Assessing correction
company to expand its product and database
service offerings. performance
With better technology infrastructure, the
company collects and stores more data, Some new tasks are created, such as establishing
builds and tests new features more and aligning data privacy and security protocols
quickly, and provides its customers with and developing for edge computing that enables
more personalized service. real-time decision making.

Source: McKinsey Global Institute analysis

McKinsey & Company

McKinsey Global Institute | Generative AI and the future of work in America 55

Filling the jobs of the future is an opportunity
to build a more inclusive economy
The challenge facing the US labor market through the end of this decade will be helping people
move out of declining occupations, equipping them with new skills, and matching them with
better opportunities. Filling the jobs with rising demand may require companies to recruit from
different talent pools—and it creates an opportunity to engage underrepresented groups.76 One
way to do this is creating new options for work that meet the needs of workers, whether that
means offering more flexible or part-time roles or hiring freelancers for appropriate tasks.
Remove unnecessary credential requirements
We estimate that workers without four-year college degrees will face 75 percent of all job
transitions that could occur by 2030. Many of these workers have what it takes to perform
higher-wage jobs, but they are often passed over. In fact, Americans without bachelor’s
degrees tend to have lower labor participation overall (58 percent versus 73 percent for college
graduates), perhaps due in part to discouragement over what kinds of roles they can land. This is
an enormous population; almost 37 percent of Americans over age 25 have a high school diploma
or less. An additional 15 percent attended college but did not earn degrees.77
To help displaced workers find new career paths, more employers will need to begin looking
beyond educational credentials to consider candidates’ transferable skills, mindset, and
potential. Companies such as IBM and Hilton are among many that are no longer requiring four-
year degrees during hiring. Pennsylvania recently removed college degrees as a prerequisite for
more than 90 percent of all state government jobs.78
A number of coalitions are working to broaden opportunity for those without college degrees.
A national campaign to “tear the paper ceiling,” sponsored by Opportunity@Work, aims to raise
awareness among employers that talent can come from community colleges, workforce training,
boot camps or certificate programs, military service, or on-the-job learning. Year Up offers high
school graduates training, six-month paid corporate internships, and networking opportunities.
The Rework America Alliance brings together employers, labor unions, community organizations,
educators, and advocacy groups to expand opportunities for workers in lower-wage occupations
to move into better jobs based on their skills and experience. One solution, still in the early
stages, involves digital learning and employment records—a sort of digital microcredential that
can document how an individual worker has acquired skills and also translate across companies
and over time.79
Remove barriers for women
It took three full years for the number of working women in the United States to fully bounce back
after the pandemic. What could be done to maintain their participation?
— Expand access to affordable, high-quality childcare. The lack of affordable childcare has
long been a major issue for parents in the United States, and it came to a head during the
pandemic. 80 In a recent McKinsey survey, women cited childcare expenses or availability as
the number one reason for voluntarily leaving the workforce. 81 Coming out of the pandemic,
more employers are adding childcare benefits, from on-site services to subsidies. But
individual employers cannot solve this issue on a large enough scale to move the needle on
women’s labor force participation. Efforts to secure federal funding for universal childcare
have not come to fruition, but some individual cities and states are acting on their own. 82

76
Aaron De Smet, Bonnie Dowling, Bryan Hancock, and Bill Schaninger, “The Great Attrition is making hiring harder. Are you
searching the right talent pools?” McKinsey Quarterly, July 2022.
77
Educational attainment in the United States: 2021, US Census Bureau, 2022.
78
Jane Thier, “Pennsylvania is axing its college degree requirement for 65,000 state jobs, calling it an ‘arbitrary requirement,’”
Fortune, January 20, 2023.
79
Learning and employment records: Progress and the path forward, American Workforce Policy Advisory Board, Digital
Infrastructure Working Group, September 2020.
80
Ryan Nunn and Katherine Lim, “A COVID-19 labor force legacy: The drop in dual-worker families,” Brookings Institution,
September 2022.
81
Sarah Gitlin, Ayushi Gummadi, Alexis Krivkovich, and Kunal Modi, “The childcare conundrum: How can companies ease
working parents’ return to the office?” McKinsey & Company, May 2022.
82
Lisa Dodson and Amanda Freeman, “The road to universal childcare will be local,” The Hill, December 23, 2022.

McKinsey Global Institute | Generative AI and the future of work in America 56

 The US infrastructure law, which requires private companies receiving contracts to provide
childcare for their workers, helps to address childcare gaps and increase the low share of
construction workers who are women; this could provide a useful template. Demand for
childcare is high, but childcare providers are among the lowest-paid workers in the US
economy—a situation that prompted Treasury Secretary Janet Yellen to comment that
childcare is “a textbook example of a broken market.”83
— Companies can reexamine hiring. Hiring managers may need to rid themselves of biases—
especially those against candidates who have gaps in their experience due to time taken for
family responsibilities. In general, companies may need to be more open to hiring candidates
with potential and training them on the job, especially when it comes to new technology skills.
— Consider “returnship” programs. JPMorgan Chase, for example, has introduced a paid
fellowship for women returning to the industry after time off. Companies with returnships and
formal mentorship and sponsorship programs have managed to develop and retain talented
women in addition to boosting their representation in the C-suite. 84
Overall, the United States could benefit from holistically considering how to boost women’s
participation in the labor market.
Keep experienced older workers engaged
The end of this decade marks a demographic inflection point. By 2030, the entire baby boom
generation will have passed age 65—meaning one in four Americans could be of retirement age. 85
As described in chapter 1, the pandemic prompted many older workers to retire early. But as
pandemic-related health concerns ease and remote work becomes more accepted, this trend
could reverse to some degree. More Americans will need to work longer to shore up their
finances, given the decline of defined-benefit pension plans. In fact, in a recent Federal Reserve
survey, only 40 percent of non-retired Americans reported that their retirement savings were on
track, and a quarter reported having no retirement savings at all. 86 In addition, Social Security is
gradually raising the age for receiving full benefits.
In a recent CNBC economic survey, 68 percent of people who retired during the pandemic
said they would consider returning to work. 87 The survey also found that retired respondents
prioritized flexible hours even above pay. Businesses that need to retain experienced workers
with hard-to-find skills may need to redesign work arrangements, perhaps focusing on
mentoring younger coworkers. Unilever’s “U-Work” model aims to provide flexibility by offering
varying intermittent assignments and allowing workers to do other things between assignments
while receiving a monthly retainer and other tailored benefits.

Helping people in low-wage,

shrinking roles add the skills they
need to move into growing jobs
will require significant effort.
83
Heather Long, “‘The pay is absolute crap’: Child-care workers are quitting rapidly, a red flag for the economy,” Washington
Post, September 19, 2021. See also Maureen Coffey, “Still underpaid and unequal: Early childhood educators face low pay and
a worsening wage gap,” Center for American Progress, July 2022.
84
Women in the workplace 2019, McKinsey & Company and LeanIn.org, October 2019.
85
Jonathan Vespa, David M. Armstrong, and Lauren Medina, Demographic turning points for the United States: Population
projections for 2020 to 2060, US Census Bureau, report P25-1144, February 2020.
86
Report on the economic well-being of US households in 2021, Federal Reserve, May 2022.
87
Steve Liesman, “Many who lost jobs during pandemic would return for the right pay and position, CNBC survey finds,” CNBC,
June 8, 2022.

McKinsey Global Institute | Generative AI and the future of work in America 57

Businesses can team up with nonprofits and local economic development agencies to find and
train experienced workers interested in “second acts.” AARP’s Back to Work 50+ program teams
with community partners to help older workers train for positions that are in demand and update
their job-hunting strategies.
Activate historically overlooked populations
Another way to increase inclusion in the labor force is by reducing barriers for often-overlooked
populations. This is one area where companies can take the lead simply by revising their own
hiring practices to take a broader perspective on candidates.
Remote work, for example, is opening up long-needed opportunities for people with disabilities,
a notably underutilized pool of talent. Employment among this group is higher than before the
pandemic. 88 Continuing this trend could be pivotal for those with chronic conditions and those
suffering from long COVID.
Residents of rural areas, too, can benefit from the shift to remote work. These areas have
historically missed out on the kind of job growth enjoyed by large coastal cities, particularly when
it comes to higher-paying and more knowledge-intensive jobs. But during the pandemic, many
urban residents relocated to smaller towns while keeping their jobs and working online. Many
companies no longer have to be constrained by geography in hiring—and there is a great deal of
untapped talent to be found in small towns and rural areas. To make the most of this opportunity,
local governments need to ensure that robust digital networks are in place.
Other traditional hiring barriers could also be eliminated, such as biases against workers with
employment gaps. Automated screening tools may be weeding out talented applicants with gaps
and other anomalies on their résumés. 89
Similarly, some employers are revisiting the requirement that employees not have criminal
records. Companies across sectors (including JPMorgan Chase and Amazon) are willing to
consider some former offenders on a case-by-case basis.90 One organization focused on making
reentry possible is Homeboy Industries. Every year, Homeboy takes hundreds of former gang
members and formerly incarcerated people through an intensive 18-month employment program,
giving them opportunities to add skills and work experience in social enterprises. Removing the
stigma surrounding those who have served their time would create opportunities for people to
redirect their lives and become contributing members of their communities.

American workers will need opportunities to

acquire the skills that will be in demand
With millions of jobs potentially being eliminated by automation—and an even larger number
being created in fields requiring different skills—the United States faces an immense challenge
and a huge opportunity. Many initiatives are in place, but not necessarily at the scale that will be
required for a smooth transition in the future. It will be critical to scale up what works and connect
displaced workers with the resources they need to make successful career transitions. The
Bipartisan Infrastructure Law, for example, includes programs focused on expanding access to
broadband as well as digital skills training, classroom instruction, and on-the-job training in fields
related to infrastructure and energy efficiency. Similarly, the CHIPS and Science Act will fund not
only semiconductor manufacturing but also STEM workforce development.
Employers can make a meaningful difference in filling roles and solving job mismatches through
new recruiting policies as well as paid internships, apprenticeships, and training programs so that

88
Don Lee, “Surge in remote working due to COVID fuels record employment for people with disabilities,” Los Angeles Times,
December 15, 2022; and Gleb Tsipursky, “Disabled people have been demanding remote work for decades. Here’s what
happened when the pandemic made it possible,” Fortune, January 3, 2023.
89
Kathryn Dill, “Companies need more workers. Why do they reject millions of resumes?” Wall Street Journal, September 4,
2021.
90
For a list, see Eugene Casey, “Companies that hire felons,” NCESC.com.

McKinsey Global Institute | Generative AI and the future of work in America 58

 individuals can earn while they learn. It will take coalitions of employers, educational institutions,
nonprofits, labor unions, and industry groups to tackle this challenge.
— Pre-college students. Initiatives in schools can prepare tomorrow’s workforce for the
science, technology, healthcare, and engineering jobs of the future. These can include
extracurricular activities and external competitions to encourage early engagement. For
example, Million Girls Moonshot runs after-school activities and summer programs to inspire
and prepare more girls to pursue studies and careers in STEM.
— College students. Many companies are teaming up with universities and community
colleges to build highly targeted vocational pipelines and academic programs in areas of
high demand.91 Educational institutions can shape curricula toward skills and growing local
industries, while companies can offer hands-on experience and immediate employment after
students complete their programs. With guidance from nonprofits and industry advisers, the
City University of New York, for example, has been revising its curriculum in data analytics
and cybersecurity to prepare students to better meet labor demand.92 Looking beyond
individual programming choices, the United States also needs to tackle some broader
structural issues, including access to education as well as rising costs and debt burdens. In
addition, institutions need to do more to boost completion rates: only about 60 percent of
students who start a four-year program obtain a bachelor’s degree, and about 30 percent of
those who pursue a certificate or associate degree actually earn those credentials.93
— Unemployed individuals. Organizations can build out training programs and pathways
to pull the unemployed into high-demand lines of work. For example, a nonprofit initiative
called Generation, founded by organizations including McKinsey, targets unemployed youth
worldwide. The program works with employers to identify the essential skills new employees
need, then distills those skills into short-term four- to 12-week technical training courses,
accompanied by mentorship and placement services with partnering employers.94
— Displaced workers. In the years ahead, millions of midcareer workers will need to leave
declining occupations and enter new arenas. Workers will need to be adaptable and rise to
the challenge of acquiring new skills. Companies can recruit from specific industries that
are shrinking, particularly if they map out which skills are transferable, and hire and train
people with potential. They can also invest in learning programs to help their own workers
make internal transfers into growing roles. With labor in shorter supply, companies will not
only need to attract new talent. They will also need to focus on retention by keeping current
workers engaged. Some attrition is to be expected, but employers cannot afford excessive
turnover and the loss of too much experience and institutional knowledge. Workers value
flexibility and a positive experience with their managers and teams more than ever—and they
are demanding more from their employers.

Workers value flexibility and a

positive experience with their
managers and teams more than ever.
91
David Tobenkin, “Employers partner with community colleges to fill the talent pipeline,” SHRM, November 2021; and Olivia
Sanchez, “Business partnerships with community colleges help funnel workers into better jobs,” Hechinger Report, July
2021.
92
Lilah Burke, “An industry-aligned education,” Inside Higher Ed, December 13, 2019.
93
“Increase completion and reduce inequities,” in The future of undergraduate education, the future of America, Commission
on the Future of Undergraduate Education, American Academy of Arts & Sciences, November 2017.
94
See generation.org.

McKinsey Global Institute | Generative AI and the future of work in America 59

Retention will also matter more than ever. Employers can use what they learned from the
pandemic to keep their workforces more engaged. In a McKinsey survey, employees cited not
feeling valued by their organizations or their managers as primary reasons for quitting. Those
who identified as non-White or multiracial were more likely than their White counterparts to say
they had left companies because they didn’t feel they belonged. Employers have learned that
employees want meaningful work, job flexibility and autonomy, connection, and opportunities
to learn and advance. Part of keeping them involves removing toxic managers in favor of those
who motivate and inspire; aligning benefits with employee priorities; strengthening the sense of
community; and creating career paths and development opportunities.95
Where it’s feasible, employers could consider hybrid work models as the rule and not the
exception. This could include both work-from-home accommodations and flexible working hours.
This is easier in occupations with clearer productivity measures. As the pandemic demonstrated,
hybrid models help sustain productivity in the face of disruptive events; they can also help
employees balance their work and personal lives. If flexibility is offered universally, but with
some clear structures and policies in place, it can be a powerful source of employee satisfaction
and loyalty.96
Offering existing employees learning opportunities and a clear path toward moving into new roles
can reduce attrition—a major cost and a perennial headache in industries such as hospitality and
retail. Walmart’s Live Better U, for example, offers employees the chance to add skills for future
roles (or even to stay in their current roles); this program covers the full cost of tuition and books
for those pursuing college degrees.
Companies can also think more broadly about where their workforce is located. Rural and low-
growth areas have lagged behind the national average in employment growth and recovery.
But remote work presents opportunities for non-urban areas to draw consumers and workers.
Not only can rural areas thrive and become more attractive to workers, but they may invest
in becoming centers for growing areas of demand such as warehousing and automation.97
Some low-growth and rural areas have been offering financial incentives such as paying
eligible workers to relocate. Other programs offer to reimburse moving expenses, provide free
recreational services, or host coworking spaces.

Some people look at the spread of automation with a sense of foreboding about what it will mean
for livelihoods. The good news is that the United States will likely have more jobs in the future,
and they will be higher-paying positions on average. But helping people in low-wage, shrinking
roles add the skills they need to move into growing jobs will require significant effort. In addition,
employers will need to redesign many jobs and make them more fulfilling, and workers will need
to adapt as technology changes the nature of what they do.
The US labor market is proving itself to be resilient, which bodes well for the nation’s ability to
handle the churn and change that the next several years may bring. Employers feel a new sense
of urgency, not only about their own workforce needs but also about what lies ahead for the US
economy. This is a moment when smart policy can make a difference—and when companies can
step up and lead in positioning people and communities for success.

95
Aaron De Smet, Bonnie Dowling, Marino Mugayar-Baldocchi, and Bill Schaninger, “‘Great Attrition’ or ‘Great Attraction’? The
choice is yours,” McKinsey Quarterly, September 2021.
96
Ellen Ernst Kossek, Patricia Gettings, and Kaumudi Misra, “The future of flexibility at work,” Harvard Business Review,
September 2021.
97
Mike Kerlin, Neil O’Farrell, Rachel Riley, and Rachel Schaff, “Rural rising: Economic development strategies for America’s
heartland,” McKinsey & Company, March 2022.

McKinsey Global Institute | Generative AI and the future of work in America 60

© People Images/Getty
Methodology brief

This research draws on a large body of MGI research on the future of work.98 Its foundational
methodology, approach, and assumptions remain consistent with those reports, with some
additions and refinements. In this report, our analysis focuses on 2022–30, and all the driver
models have been updated accordingly. Additionally, we have updated two of our trends to
incorporate developments that have gained momentum in the United States, as follows:
— The push to achieve net-zero emissions: The United States has made a formal international
commitment to reduce greenhouse-gas emissions to net zero by 2050 at the latest. We
examined the impact of this move on jobs using a scenario-based analysis drawing on the
approach used by the Network for Greening the Financial System. We built on previous
McKinsey research assessing the impact at a global level to analyze the specific implications
for the United States, adding sectoral and occupational dimensions to the analysis.99
This net-zero analysis assumes that current supply chain composition remains the same by
2030. We consider the impact of the United States meeting its own emissions targets as well
as its role in meeting global demand for products. Jobs gained and lost are allocated as per
the occupational mix of 2022. We consider job losses and gains directly associated with the
transition and do not include other macroeconomic forces like population and income growth.
In addition to creating a scenario, we consulted other estimates of the net-zero transition’s
impact on jobs by 2030. These range from job gains of 178,000 (Decarb America) and
300,000 to 600,000 (Princeton Net-Zero America) to gains of 6.5 million (World Resources
Institute 2035 estimate).100 Estimates vary depending on assumptions about productivity and
the US role in the global energy market, plus whether direct and indirect impacts are included.
— Infrastructure spending: The Bipartisan Infrastructure Law, signed in 2021, allocated
$1.2 trillion in federal funding toward a variety of projects nationwide.101 To reflect the
impact of additional spending on jobs creation mandated by the legislation, we updated our
infrastructure driver by incorporating construction-sector value-added estimates inclusive of
Bipartisan Infrastructure Law allocations through 2030.
We use the following data sources and methodology:
— Labor force data: The Current Population Survey (CPS), conducted by the US Census Bureau
for the US Bureau of Labor Statistics (BLS), is the primary data source for workforce statistics
in this report. The CPS provides monthly data on detailed labor force variables, which are
used here as reported at the source. The Census Bureau integrates updated population
estimates into the CPS each January, but the historical estimates are not revised. We use
the published unrevised estimates for purposes of comparison, while acknowledging the
data limitations.

98
This research draws on the methodology and findings of multiple MGI reports: A future that works: Automation, employment,
and productivity, January 2017; Jobs lost, jobs gained: Workforce transitions in a time of automation, November 2017;
The future of work in America: People and places, today and tomorrow, July 2019; and The future of work after COVID‑19,
February 2021.
99
The net-zero transition: What it would cost, what it could bring, McKinsey Global Institute, January 2022.
100
Employment impacts in a decarbonized economy, Decarb America, June 2022; Net-zero America: Potential pathways,
infrastructure, and impacts, Princeton University, October 2021; and How a clean energy economy can create millions of jobs
in the US, World Resources Institute, September 2022.
101
“President Biden’s Bipartisan Infrastructure Law,” The White House, November 2021.

McKinsey Global Institute | Generative AI and the future of work in America 63

 Box 2
A note on terminology
We use the term “occupational shifts” to mean declines in employment in specific
occupations between 2019 and 2022. (We do not include gains in this calculation to avoid
double counting.)
Since the number of total workers grew over this period, losses in some occupations
were offset or more than offset by gains in other occupations. We interpret these losses
or declines that are offset by other gains as the effect of workers switching from one
type of work to another. From the gross declines in each occupation, we also reduced a
proportional share of total retirements. We calculated this by taking the number of total
retirements from the BLS and allocating a share to each occupation according to its
proportion of workers ages 65 and over.
However, we do not know exactly how many individuals moved from one occupation
to another or if they made multiple moves; for that reason, we refer to the number of
occupational shifts that occurred during this period rather than specifying the number of
workers making those changes.
Much of our analysis looks at occupational categories, using BLS groupings. Each of
those categories contains multiple occupations. A move from one occupation to another
within the same category is counted as a shift (in historical context) or a transition (in the
context of future projections). Within any given category, some individual occupations
may experience declines while others increase. The results by category reflect their
offsetting effects.

— Baseline employment: We use Occupational Employment Statistics data for 2022 from the
BLS to capture the occupational structure of the workforce across sectors. The numbers are
then proportionately scaled to CPS total employment to enable the inclusion of farm and self-
employed workforce statistics. We apply annual growth rates from BLS 2021–31 projections
to the 2022 baseline to estimate 2030 employment levels.
— Automation: To determine the potential impact of automation in the workplace, we use data
published by the BLS and O*NET to break down some 850 occupations into about 2,100
constituent activities. Each activity is matched to one or more of 18 capabilities to assess
the potential for a known technology to handle this work. In addition to looking at what is
possible, we also consider what is feasible by assessing the solution development timeline,
the economics of adoption versus paying wages, and the time required for adoption and
deployment. We use the scenario inclusive of the impact of generative AI.
— Impact of automation on productivity: We used GDP per full-time employee (FTE) as the
measure of productivity. We calculated automation output under different scenarios by
multiplying the projected number of FTEs by the estimated automation adoption rate. To
maintain consistency with other data sources, we made several additional assumptions. We
considered only job activities that are available and well defined as of the date of this report.
Also, to be conservative, we assumed automation has a labor substitution effect but no
other performance gains. Finally, we created two scenarios: a pessimistic scenario in which
labor displaced by automation rejoins the workforce at 2022 productivity levels, and a more
optimistic scenario in which they rejoin at 2030 productivity levels, net of automation. In both
scenarios, we have incorporated labor displaced rejoining in line with the expected 2030

McKinsey Global Institute | Generative AI and the future of work in America 64

 occupational mix. We also assumed that additional output from automation will not decrease,
even if the total number of FTEs declines as a result of demographic changes. Under the
assumptions outlined above, we calculated the GDP impact of automation adoption by
country as follows:
Additional GDP impact of automation = FTE impact of automation X productivity2022
The additional GDP impact of automation is then added to 2022 GDP to estimate the productivity
impact of automation.
— Long-term labor market trends: We estimate labor demand factoring in seven
macroeconomic catalysts: rising incomes, healthcare and aging, technological innovation,
infrastructure investment, education, the marketization of unpaid work, and the push to reach
net-zero emissions across the economy. We capture direct and indirect jobs that could be
created or lost from each catalyst by leveraging job multipliers from input-output tables.
— Trends accelerated by COVID-19: We incorporated several broad trends accelerated by
the pandemic that may influence labor demand and jobs in the economy through 2030,
including increased remote work and virtual meetings and the shift to e-commerce and other
virtual transactions.
— Skills demand and transitions: We look at 25 workforce skills that fall into five broad
categories: physical and manual, basic cognitive, higher cognitive, social and emotional, and
technological. The number of hours that workers spend performing specific work activities
are then mapped to each skill. Since one-to-one mapping of skills and activities undercounts
digital skills, we correct this by reallocating a portion of hours from activities requiring non-
technological skills to those requiring basic digital skills.

McKinsey Global Institute | Generative AI and the future of work in America 65

© Eclipse Images / Getty
Acknowledgments
This report is the product of a collaboration between the McKinsey Global Institute and the
McKinsey Center for Government.
The research was led by Kweilin Ellingrud, an MGI director based in Minneapolis; Saurabh
Sanghvi, a McKinsey partner based in Silicon Valley; Gurneet Singh Dandona, an MGI senior
expert and associate partner based in New York; Anu Madgavkar, an MGI partner based in New
Jersey; Michael Chui, an MGI partner based in San Francisco; and Olivia White, an MGI director in
San Francisco. Paige Hasebe, Alok Singh, and Dian Yu led the project team, which included Juhi
Daga, Maclaine Fields, Klaudia Gegotek, Darien Ghersinich, Megan Goldring, Segun Omole, and
Karis Tai. This report was edited and produced by MGI executive editor Lisa Renaud, as well as
editorial operations manager Vasudha Gupta and senior data visualization editor Chuck Burke.
Additional research support was provided by Tim Bacon, Arthur Bianchi, Luis Campos, Kanmani
Chockalingam, Kathryn Fox, Ian Murphy, Jose Maria Quiros, Prit Ranjan, and Soyoko Umeno.
Many McKinsey colleagues provided valuable insights and expertise. We thank Brendan
Buescher, Eric Chewning, Trey Childress, Rory Clune, André Dua, Drew Erdman, Anna Fu,
Steffen Fuchs, David Fuller, Bryan Hancock, Garo Hovnanian, Mike Kerlin, Mekala Krishnan, Tom
Latkovic, Nicholas Laverty, Ryan Luby, Elizabeth Murthy, Fiyin Olaridan, Ron Ritter, Rafael Rivera,
Ben Safran, Navjot Singh, Steven Smith, Sarah Tucker-Ray, and Jonathan Woetzel.
We are grateful to our colleagues Sharmeen Alam, Laura Barbara, David Batcheck, Tim Beacom,
Fiorella Castello, Amanda Covington, Alexandra Dorn, Christina Fisher, Stephanie Jones, Rebeca
Robboy, Rachel Robinson, Patricia Sabga, Ben Saft, Jesse Salazar, Stephanie Strom, and Nathan
Wilson for their support and contributions.
This research contributes to our mission to help business and policy leaders understand the
forces transforming the global economy. As with all MGI research, it is independent and has not
been commissioned or sponsored in any way by any business, government, or other institution.
We welcome your comments on this research at [email protected].

McKinsey Global Institute | Generative AI and the future of work in America 67

McKinsey Global Institute
July 2023
Copyright © McKinsey & Company
Designed by the McKinsey Global Institute
www.mckinsey.com/mgi
@McKinsey_MGI
@McKinseyGlobalInstitute
@McKinseyGlobalInstitute
Subscribe to MGI’s podcast, Forward Thinking:
mck.co/forwardthinking