1.

Topic introduction

Machine learning (ML) is a discipline of artificial intelligence (AI) that provides machines with
the ability to automatically learn from data and past experiences while identifying patterns to
make predictions with minimal human intervention. Machine learning algorithms use historical
data as input to predict new output values. ML algorithms use computation methods to learn
directly from data instead of relying on any predetermined equation that may serve as a model.

Machine learning relies on input, such as training data or knowledge graphs, to understand
entities, domains and the connections between them. With entities defined, deep learning can
begin. The machine learning process begins with observations or data, such as examples, direct
experience or instruction. It looks for patterns in data so it can later make inferences based on the
examples provided. The primary aim of ML is to allow computers to learn autonomously without
human intervention or assistance and adjust actions accordingly. Machine learning algorithms
are molded on a training dataset to create a model. As new input data is introduced to the trained
ML algorithm, it uses the developed model to make a prediction. Further, the prediction is
checked for accuracy. Based on its accuracy, the ML algorithm is either deployed or trained
repeatedly with an augmented training dataset until the desired accuracy is achieved. Some
practical examples of machine learning include language translation, image recognition, chat
bots, and predictive analytics.

Machine learning is often categorized by how an algorithm learns to become more accurate in its
predictions. There are four basic approaches:

 Supervised learning: In this type of machine learning, data scientists supply algorithms
with labeled training data and define the variables they want the algorithm to assess for
correlations. Both the input and the output of the algorithm is specified.
 Unsupervised learning: This type of machine learning involves algorithms that train on
unlabeled data. The algorithm scans through data sets looking for any meaningful
connection. The data that algorithms train on as well as the predictions or
recommendations they output are predetermined.
 Semi-supervised learning: This approach to machine learning involves a mix of the two
preceding types. Data scientists may feed an algorithm mostly labeled training data, but
 the model is free to explore the data on its own and develop its own understanding of the
data set.
 Reinforcement learning: Data scientists typically use reinforcement learning to teach a
machine to complete a multi-step process for which there are clearly defined rules. Data
scientists program an algorithm to complete a task and give it positive or negative cues as
it works out how to complete a task. But for the most part, the algorithm decides on its
own what steps to take along the way.

2. Reason for selecting the topic

AI and machine learning are in their early stages but are evolving at a rapid pace. Machine
learning is changing many facets of the business world. According to Digital Marketing Institute,
97% of business leaders believe the future of marketing consists of marketers and machines
working in tandem to better a business's outcome.

Machine learning is an expanding field. Within the business industry, machine learning and
artificial intelligence have gained incredible popularity. More and more businesses are choosing
to utilize this type of technology, which has led to a boom in machine learning advancements.
Machine learning has become increasingly popular as, when used in the right way, it can help to
manage a number of complex business problems, as well as predicting client behavior. Today,
with the rise of big data, IoT, and ubiquitous computing, machine learning has become essential
for solving problems across numerous areas, such as:

 Computational finance (credit scoring, algorithmic trading)

The research conducted by Google cloud found that there is a big global opportunity for ML in
terms of current investment and expected future growth. An MIT Technology Review survey of
375 leading businesses in over 30 countries found that 60% of respondents had already
implemented an ML strategy and had committed to ongoing investment in ML. Global market
revenue for ML as a service (MLaaS), or vendor platforms for ML, totaled US$1.07 billion in
2016, and is expected to grow to US$20 billion by 2025.

After the ground shifting changes brought in by the extensive use of computers and smart
devices over the last few decades, artificial intelligence (AI), robotics and machine learning are
now the new tools to unlock future potentials to positively impact the lives of people the world
over. Nepal is also witnessing the unfolding of this technology revolution as some creative and
geeky techies in the country have taken charge to bring positive changes in the society. There is
an immense potential of machine learning to transform the economic situation of countries like
Nepal. Hence, it is prerequisite to have knowledge about machine learning.
3. Importance of machine learning to the business
4. Business Benefits

The market for ML platforms is heating up. According to International Data Corporation (IDC),
spending on AI and ML will grow from $12B in 2017 to $57.6B by 2021 (Deloitte, 2017).
Several leading vendors are providing cloud base ML services including Microsoft, Google,
IBM, Oracle, and Amazon. Most ML platforms span the entire ML lifecycle from data ingestion
to model development to production. These platforms are cloud-based, simple, and could get
clients started with ML allowing them to develop models that could be quickly put in production.
Many routine internal processes can be done quicker and more systematically by machines than
people. The MIT Technology Review (2017) survey of businesses found that 30% of adopters of
ML had gained internal process efficiencies.

Google Cloud surveyed 20 business leaders who have implemented ML projects in their
organizations. The survey identified key benefits yielded from the projects (Google Cloud,
2017). Top benefits included: time savings, cost savings, better risk management, improved
quality of analytics, increased revenues, improved services, improved inventory planning.

HR professionals often have to review hundreds, or (for larger businesses) even thousands, of
resumes and cover letters for open positions. Machine learning has helped with recruiting and
hiring by standardizing the process: It can analyze and organize resumes, scrape social media
channels and build candidate profiles.

ML-based system is the answer to various human labor-intensive tasks such as stock market
prediction, medical diagnosis, employee selection, and electrical demand prediction. ML offers
solutions to some of the most important challenges faced by different business sectors today
(Zheng, 2015). ML is being used today to make driver-less cars, predict emergency room waiting
times, and make intelligent recommendations on movie selection on services, such as Netflix.
ML techniques and models are often employed to learn patterns from past data and use them to
make predictions about the future. Building predictive applications using ML is becoming easier
as the next frontier, next-generation machine learning platforms strike a balance between ease-
of-use and customizability (Zheng, 2015).

A good example of machine learning is Amazon’s Alexa device. You can ask Alexa to play any
music you want. Over time, the data you’re giving Alex will help the AI determine what type of
genre you like. When you say, “Alexa, play music,” it will play music from that genre. Alexa has
spotted a pattern and will create a playlist that it thinks you will enjoy.

1Web and social media – ML can be used for intelligent typo detection and correction (ITDC)
system. ML is capable of automatically identifying and accurately correcting a broad range of
typos, from simple typos to complex spelling errors (Huang et al., 2015). Additionally, it has
been shown that ML can be used to perform main content filtering of the web page (Narwal and
Sharma, 2016). ML is being used effectively by online businesses as a recommendation engine
for shopping on the internet and to personalize online ad delivery in almost real-time. Other
common ML use cases include fraud detection, network security, threat detection, spam filtering,
predictive maintenance and building news feeds. For example, Facebook is using ML to
personalize each member’s feed. The software is using statistical analysis and predictive
analytics to identify patterns in the user’s data and use those patterns to populate the news feed.
Should a member frequently scroll to read or ‘like’ a particular friend’s posts, the news feed will
start to show more of that friend’s activity earlier in the feed. The news feed will adjust
accordingly should the member stop to read, like or comment on the friend’s posts (Rouse,
2017). Yelp is utilizing ML algorithms to compile, categorize, and label tens of millions of
photos of businesses more efficiently. Pinterest turned to ML when it acquired Kosei, a ML
company in 2015. Through the use of Kosei’s algorithms, Pinterest empowered users with the
ability to discover content and receive more accurate recommendation (D’Onfro, 2015). Lastly,
Salesforce’s CRM software is enabled with a proprietary ML technology called Einstein, which
allows businesses analyze every aspect of a customer’s relationship from initial contact to
ongoing engagement (Marvin, 2016).

Online chatbots are replacing human agents along the customer journey, changing the way we
think about customer engagement across websites and social media platforms. Chatbots answer
frequently asked questions (FAQs) about topics such as shipping, or provide personalized advice,
cross-selling products or suggesting sizes for users. Examples include virtual agents on e-
commerce sites; messaging bots, using Slack and Facebook Messenger; and tasks usually done
by virtual assistants and voice assistants. Some widespread examples of ML today include
Google’s page ranking system, photo tagging on Facebook, and automatic spam filtering on
Gmail (Analytics Vidhya, 2016).
2 Financial institutions and legal firms – in recent years, ML has impacted the banking industry
in many positive ways. Banks have benefited from better customer interaction using chatbots,
more robust and intelligent anti-money laundering and fraud detections, and accurate customer
recommendations using recommendation engines (Mangani, 2017). ML applications in finance
are well documented. Fund administrators are the primary users of ML platforms. They use ML
to review potential risks to their managed portfolio and to test different trading strategies to see
how various options would affect investment risks. Moreover, ML enables fund administrators to
assess characteristics of high-risk environment over time and quantifies the risk for investors.
This kind of platform was used during UK’s Brexit vote to minimize fallouts for investors. The
vote to leave the European Union had a major impact on financial markets. However, those
investors that used omega point portfolio intelligence platform saw little change in the value of
their portfolio after the Brexit vote. The platform uses ML models and algorithms that take into
account nearly 50 economic indicators including company specific financial reports and
macroeconomic data to review portfolios for risks (Burns, 2016). Finally, services such as legal
robot can be used to determine whether all the necessary clauses are included in business
contracts, aiding lawyers from having to spend wasteful time and resources in doing so.
Designed to optimize stock portfolios, AI-driven high-frequency trading platforms make
thousands or even millions of trades per day without human intervention. Banks and other
financial institutions can use machine learning to spot suspicious transactions. Supervised
learning can train a model using information about known fraudulent transactions. Anomaly
detection can identify transactions that look atypical and deserve further investigation.

3 Manufacturing and warehousing – manufacturers are fitting equipment with sensors and
embedding analytic sensing systems to identify potential equipment failures before they happen.
They are also using ML to improve customer service, maintenance and warranty services for
their clients (Nevala, 2017). Moreover, ML is utilized for machine vision, demand forecasting,
process optimization, and condition monitoring. In Amazon’s automated distribution centers
robotic systems collect required supplies and validate the right stuff gets in the box (Nevala,
2017). ML is also shown to have potential for continuous improvement of heterarchical
manufacturing control system (Aissani et al., 2008). Kewpie, a major food manufacturer in
Japan, uses machine learning to implement strict quality control of the diced potatoes used in
manufacturing baby food. Without machine learning, each quality control inspector would assess
more than one million diced potatoes by eye every day. Takeshi Ogino, Deputy General Manager
of Production Division believes that “machine learning has been valuable for Kewpie to achieve
this in a much more efficient manner.”

4 Retail – retailers are utilizing ML for real-time online pricing, to determine customer responses
or purchases and set up for upselling and for cross selling, and market segmentation and
targeting. They are also using ML to find fraudulent transactions over e-commerce, and analyze
customers’ sentiments on products seen on multiple websites. Finally, ML is utilized for
predicting trends, recommend new products, and boost profitability (Nevala, 2017). In February
of 2018, Amazon unveiled its newest version of echo called echo look – a new gadget with a
service that combines ML algorithms with advice from fashion specialists. It is designed with a
focus on personal style. The unit has a hands-free camera optimized to capture your outfits, and
provide style assistant where you get a second opinion on your outfit. A dedicated app helps you
to see yourself from every angle and build your personal look book (Amazon Web Site, 2018).

Using past consumption behavior data, AI algorithms can help to discover data trends that can be
used to develop more effective cross-selling strategies. This approach is used by online retailers
to make relevant product recommendations to customers during the checkout process. A key
example of this is when you go shopping at a supermarket, and when you pay for your items, a
voucher is printed out offering money off of items frequently purchased. AI technology is used
to create these personalized offers for customers, boosting feelings of brand loyalty and
contentment.

5 Healthcare – unsupervised ML models can be used in decision supportive tools for predictive
grading of autistic disorder (Pratap and Kanimozhiselvi, 2014). Data mining and ML methods
have also been developed and used for large scale healthcare data analysis (Zhang et al., 2015).
In the healthcare industry, ML may be used to predict emergency wait room times, omit
repetitive filling out of forms for admissions and readmissions, and analyze trends regarding
strokes and seizures to name a few. ML may also be used to supplement the skills of human
radiologists by identifying changes in imaging scans more quickly, leading to earlier and more
accurate diagnoses. Thanks to wearable sensors and devices, the vital statistics of patients can be
monitored in real-time as well (Bresnick, 2017).
5. Business Challenges (should be backed by literature/ Example)

Just like with any technology, businesses must carefully evaluate when and how ML should be
applied – ideally this is done deliberately and only when it can deliver a specific, measurable
business value. Most enterprises will fail to leverage the power of ML due to the lack of quality,
clean data and inherent biases within this data. Another challenges enterprise could face using
ML is loss of control. Management may not be able to keep up with the pace of the results
generated by ML, therefore careful consideration must be required for these types of
implications. There is a judgment call needing to be made as how autonomous the ML model
would be. It could initially be decided that it will be partially autonomous, with certain aspects
being run by a human workflow, before deploying it as fully autonomous, depending on the
comfort level of the business with technology. Businesses must also evaluate if they are ready
and willing to take action given the insights found. This involves re-evaluating manpower
capabilities, funding capabilities, and buy in from upper management and employees to name a
few factors. Perhaps the existing organizational structure would need to be modified to
accommodate such changes, as roles and processes are deleted, generated, or further developed
(Krishna et al., 2017).

As with the usage of any technology, there are certain risks associated with ML, for example,
hackers may be able to manipulate an algorithm, such that it provides false positive outcomes. Or
perhaps the data itself becomes contaminated or compromised by hackers, who have found
managed to find loopholes in the system running the ML application platforms. Hackers can fool
ML models to classify malicious training samples as legitimate or vice versa, resulting in
misleading outcomes, which humans may not be able to distinguish as being un-useable. With
the popularity of biometrics, especially with the recent launch of the iPhone X, there are new
opportunities for hackers to use fake biometric fingerprints, iris, and facial characteristics to
impersonate legitimate users (Krishna et al., 2017).

Data Acquisition and Storage

Data acquisition and storage is a real challenge in AI implementation. Industrial AI systems

Another challenge of artificial intelligence is that not all business owners or managers are willing
to invest in it. The funds required to set up and implement Artificial Intelligence is very high,
thus not every business owner or organization can invest in it or can try it for their own business
(Greaves, 0. et al)

Integrations

Ultimately, the success of AI implementation within an organization will depend on how well
the solution integrates with existing infrastructure and business functions. An organization needs
to be flexible enough to adapt new business models, new team models, and new workflows
across all departments and teams. Incorporating AI into the business is as much a people and
process problem as it is a technology one (Kharuf, A 2019)

Consumer Trust

People tend to feel uncomfortable when confronted by things that they do not understand. When
rep01is and evaluations are made without any human intervention, it can be difficult for
consumers to understand and trust a decision that affects them.

Infrastructure

Data handling, storage, computes, scaling, extensibility and security are all critical components
necessary for enterprises to deploy an AI solution. An organization's ultimate success with AI
always starts with how suitable its infrastrncture environment is to support powerful AI
applications and workloads (Petuum 2019)

Legal Challenges

An AI application with an erroneous algorithm and data governance can cause legal challenges
for the company. This is yet again one of the biggest Artificial Intelligence problems that a
developer faces in a real world. Flawed algorithm made with an inappropriate set of data can
leave a colossal dent in an organization's profit. An erroneous algorithm will always make
incorrect and unfavorable predictions. Problems like data breach can be a consequence of weak
& poor data governance-how? To an algorithm, a user's PII (personal identifiable information)
acts as a feedstock which may slip into the hands of hackers. Consequently, the organization will
fall into the traps of legal challenges.

Complex algorithms

The function and performance are highly dependent on algorithms and the neural networks. An
organization needs to have clarity how AI is going to impact the business outcomes before
extending its applications. Understanding the complex algorithms will reduce the legal and
regulat01y challenges to the organization (Kumari, S. 2018)

Impact on Employee Morale

Extending to the previous challenge, AI does challenge the morale of its human counterpart. It's
complicated when we are building robots and scripts to take over our jobs