See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/353435681

AI Transformation in the Public Sector: Ongoing Research

Conference Paper · July 2021

DOI: 10.1109/SAIS53221.2021.9483960

CITATIONS READS
5 437

4 authors:

Einav Peretz-Andersson Niklas Lavesson

Jönköping University Blekinge Institute of Technology
4 PUBLICATIONS 85 CITATIONS 104 PUBLICATIONS 1,324 CITATIONS

SEE PROFILE SEE PROFILE

Albert Bifet Patrick Mikalef

Télécom Paris Norwegian University of Science and Technology
312 PUBLICATIONS 19,076 CITATIONS 200 PUBLICATIONS 14,681 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Patrick Mikalef on 24 July 2021.

The user has requested enhancement of the downloaded file.

 AI Transformation in the Public Sector:
Ongoing Research
Einav Peretz-Andersson Niklas Lavesson
Department of Computing, School of Engineering Department of Computing, School of Engineering
Jönköping University, Gjuterigatan 5 Jönköping University, Gjuterigatan 5
2021 Swedish Artificial Intelligence Society Workshop (SAIS) | 978-1-6654-4236-7/21/$31.00 ©2021 IEEE | DOI: 10.1109/SAIS53221.2021.9483960

SE–553 18 Jönköping, Sweden SE–553 18 Jönköping, Sweden

Email: [email protected] Email: [email protected]

Albert Bifet Patrick Mikalef

Te Ipu o te Mahara AI Institute Department of Technology Management
University of Waikato, Gate 1, Knighton Road SINTEF Digital, Strindvegen 4
Hamilton 3240, Waikato, New Zealand 7034 Trondheim, Norway
Email: [email protected] Email: [email protected]

Abstract—Real-world application of data-driven and intelligent data-driven and intelligent systems (AI) in the private and
systems (AI) is increasing in the private and public sector as public sectors as well as in society at large. As was the case
well as in society at large. Many organizations transform as a in earlier societal revolutions, old systems, schools of thought,
consequence of increased AI implementation. The consequences
of such transformations may include new recruitment plans, organizations, jobs, and cultures will change or disappear.
procurement of additional IT, changes in existing positions However, we argue that the AI revolution – the shifting of
and roles, new business models, as well as new policies and the cognitive workload from humans to computers – may
regulations. However, it is unclear how this transformation varies have characteristics and consequences quite unlike the earlier
across different types of organizations. We study the effects of revolutions.
bottom-up approaches, such as pilot projects and mentoring to
specific groups within organizations, and aim to explore how We are interested in understanding how and why organi-
such approaches can complement the top-down approach of zations are transformed as a consequence of AI introduc-
strategic AI implementation. Our context is the public sector. tion, implementation, and increased readiness. We believe the
Our goal is to acquire an improved understanding of how and strategic planning, organizational culture, and decision mak-
when AI transformation occurs in the public sector, which are the
ing processes differ significantly between typical public and
consequences, and which strategies are fruitful or detrimental to
the organization. We aim to study public sector organizations in private sector organizations. The most prominent explorations
Sweden, Norway, New Zealand, Germany, and The Netherlands of AI transformation have arguably been carried out by and
to learn about potential similarities and differences with regard for the private sector. In particular, the focus has been on the
to AI transformation. information technology industry. Two notable developments in
this respect represent fitting illustrations:
I. I NTRODUCTION
• Landing.AI: Andrew Ng – founder of Coursera – re-
The digital transformation of society has been steadily leased the AI Transformation Playbook1 on December
increasing during the last five decades. The Internet era of 14, 2018. The playbook draws from experience of AI
the last two decades has boosted this development and it has transformation from the Google Brain team and the
made possible the information sharing between individuals and Baidu AI Group. The purpose of the playbook is to
organizations, and across nations, that we experience today. deliver insights to organizations that can help them to
The early industrialization, later industrialization, and the transform into strong AI actors. It specifically addresses
digital transformation along with the introduction of renewable large private enterprises.
energy sources, are referred to as the first, second, and third • ML-Ops: In 2015, a paper about the technical debt
industrial revolution. We are now entering the era of the fourth of ML systems was published at the influential Neural
industrial revolution. Information Processing Systems (NIPS) conference [1].
The advent of social media and electronic commerce plat- The paper, which was written by a number of Google
forms, such as Amazon, Facebook, and Twitter – along with employees, reported that it was common to incur massive
the development of cheaper and more powerful hardware – ongoing maintenance costs in real-world ML systems.
has significantly increased real-world application of connected, As a consequence, initiatives such as ML-Ops have

1 AI Transformation Playbook,
978-1-6654-4236-7/21/$31.00 ©2021 IEEE https://landing.ai/ai-transformation-playbook/

Authorized licensed use limited to: Norges Teknisk-Naturvitenskapelige Universitet. Downloaded on July 24,2021 at 16:07:25 UTC from IEEE Xplore. Restrictions apply.
 been started to establish effective practices and processes have an impact on human lives should communicate in
around designing, building, and deploying ML models a transparent, fair, and accountable way [4].
into production2 . 2) Accountability and Responsibility. The legal issues
It is conceivable that the rules of the AI playbook and the concerning AI requires more discussion to determine
guidelines and principles of ML-Ops can be adapted suc- who should be responsible for the consequences of AI
cessfully to suit small and medium sized private enterprises. actions and behavior, and to define rules and guidelines
Similarly, the rules and guidelines may provide insights that for privacy, safety, and integrity.
could benefit non-information technology actors or public 3) Economy of Scale: AI may create imbalances in the
sector stakeholders. The question is which rules and guidelines economy of scale or disrupt small companies [5]. This
will apply in such contexts, and to what extent. Another is a situation that can lead to the increased monopoly of
question is whether insights gained from AI transformation large organizations.
in the public sector carries across nations the same way such
insights would from AI transformation in the private sector. The introduction of smart technologies creates ambiguity and
uncertainty. AI may cause destructive creation in the future
II. O UTLINE organization. However, from the nature of capitalism it may
This abstract seeks to sketch and motivate our long-term also create paths to growth [6].
AI transformation research agenda and international collabo- According to the Cambridge Dictionary, transformation
ration, and to describe an ongoing case study of bottom-up AI refers to a complete change in the appearance or character of
transformation approaches in the Swedish public sector. something or someone, especially so that that thing or person
We first provide a brief background on organizational is improved. However, in the area of business administration,
change, digital transformation, and the relationship to AI transformation usually refers either to radical change or to
implementation. incremental change in organizations. Organizational Transfor-
mation (OT) has been discussed in the organization and man-
III. BACKGROUND agement literature from various perspectives. It has been con-
Organizations change and develop their business models in sidered as a form of radical change [7]–[9], strategic change
relation to digitalization. Organizations are operating in a com- [10], revolutionary change [11], continuous and confluent
plex environment where new technologies require continuous organizational change [12], and organizational discontinuous
change. The introduction of AI represents one important type change [13], [14]. The discussion in the literature about OT
of change. The new age of AI will change the composition, dispels ambiguity. However, we consider OT as a consequence
business models, and tasks required in an organization. New of changes happening due to AI technology use.
business models can be a result of strategy or a strategizing In our view, AI transformation is an interdisciplinary re-
action [2]. AI significantly changes the composition of the search topic, and we believe that researchers should opt for
resources, operations, and structure that an organization can new, innovative approaches when exploring this phenomenon.
employ in order to become more efficient and to create value. The diverse aspects of the topic can be psychological, socio-
AI is viewed by many as a revolutionary and societal change cognitive, socio-technical, economic, and political. Research
agent, which will affect organizational strategy. It will involve on the topic therefore needs to consider the relevant aspects
economical, psychological, technological, political, and ethical for the context [15].
aspects. This change will directly and indirectly affect the
business models, including the purpose, process, strategies, In order to strategize AI, it is important to quantify and
infrastructure, organizational structures, and operational pro- understand the organization’s AI maturity, since maturity is a
cesses and policies. measure that relates to the organizational readiness and AI
The theory of dynamic capabilities implies that organiza- capability. The organizational inertia is a vital aspect that
tions are experiencing constant change, which force them to makes OT an important theoretical and practical problem [16].
adapt, integrate, and reconfigure their internal and external In terms of AI-enabled OT research, the existence of orga-
competences in order to maintain their competitive advan- nizational inertia raises the question where the focus should
tage [3]. be put: Which are the relevant aspects and which unit of
AI technologies affect regulations and policies. From an analysis should be considered? (individual vs. departmental,
institutional point of view, some aspects need to receive special organization vs. sector, and so forth) [16].
consideration: We are unable to find much empirical research concerning
1) Governance. Governance concerns the responsibility for AI-related OT in the public sector. Meanwhile, public sector
the risks involved in AI technologies. Governance should companies and authorities face a multitude of challenges re-
be the bridge of information and knowledge to society in lated to digital and AI transformation of society. We therefore
regard to AI. It should ensure that AI technologies that argue that AI-related OT deserves more attention from the re-
search community. This attention needs to be interdisciplinary
2 ML-Ops, https://ml-ops.org in nature.

Authorized licensed use limited to: Norges Teknisk-Naturvitenskapelige Universitet. Downloaded on July 24,2021 at 16:07:25 UTC from IEEE Xplore. Restrictions apply.
 A. Related Work B. Systematic Literature Review
We are unable to find a concise and useful definition of We have conducted a systematic literature review [17]. The
AI transformation [17]. The available research that brings aim of the SLR was to aggregate the body of knowledge on
up this phenomenon is often focusing on digital transfor- the relationship between AI and organizational transformation,
mation. There is a substantial scientific discussion around to map the field, and to identify the gaps in research that
digital transformation but very few works focus on AI. The represent an opportunity for future study. We have used three
digital transformation phenomenon refers to the relationship databases (Science Direct, Scopus, and IEEE) and found 966
between three domains: data, digital technology, and peo- papers related to the topic. We have followed the procedure
ple [18]. Transformation of an organization or a network and guidelines of Kitchenham [22] and narrowed the article
of organizations occurs in a variety of contexts, such as: selection to a final number of 52 relevant articles, which help
cultural, technological, and governance strategy [19], Various us to explore AI transformation.
definitions related to digital transformation are presented [20],
however, we observe the need to provide a definition of AI C. Interviews with Jönköping Municipality
transformation specifically. The reason for this is that, unlike We have performed interviews with 23 officials from the
other forms of digital transformation, AI will shift cognitive official management of Jönköping Municipality. The purpose
work from human actors to computers. The consequences for is to understand how AI implementation transforms public
many organizations will be significant. Organizations often sector organizations. The municipal organization represents an
recognize the need to implement AI as part of their vision interesting context. The aim is to contribute to the knowledge
and strategy but training and tutoring regarding AI and its of how the municipal board and civil servant management and
capabilities must be a first anchor in all levels in the organiza- other management groups create, formulate, and implement
tion. AI is affecting organizations in at least two directions. It digitization and AI strategies. A special focus will be put on
results in increased digitization of the economy and it enables how AI can contribute to the municipality’s work to improve
the automation of existing processes [21]. In addition, AI has the organization’s efficiency.
a disparate effect on organizations: on the one hand it creates
promising opportunities for the future, but on the other hand D. Bottom-up AI Transformation
it involves challenges, uncertainties, and risks. Defining and
explaining AI transformation is therefore crucial, since a useful We have asked Jönköping Municipality to define use cases
definition will support organizations to adapt to AI. or challenges within its respective branches and municipal
companies. We have then supported the municipality in devel-
oping the most mature use cases into research and development
IV. O NGOING P ROJECTS projects. We are now assigning students and researchers to
develop prototypes and experiments related to four of the use
This section describes ongoing projects and collaborations
cases:
regarding various aspects of AI transformation research and
real-world state-of-practice. UC1 Contact Center. The goal is to learn from a data set
of historical incoming citizen requests and responses to
generate a decision support system that can recommend
A. AI Strategy Development
categorizations of requests and formulation of responses
AI is an important tool to handle strategic issues which human operators.
also affects the ability of the organization to achieve strategic UC2 Technical Office. The goal is to learn from a data set of
change [2]. AI can be both a narrow, specific strategic issue historical citizen incident reports to generate a model that
or broad and general. For example, if AI is viewed as a can recommend actions to human operators (for example,
technology, a strategic issue might be to procure an AI system. to file a police report or provide a specific response).
However, if AI is viewed as a revolutionary and societal UC3 Technical Office. The goal is to learn from a data set of
change agent, the strategic issue will be broad and involve historical citizen incident reports about winter weather
economical, psychological, technological, political, and ethical consequences to generate a model that can formulate
aspects. We believe it is important to learn about various automated answers to citizens and prioritize planned
organizations’ strategic planning processes and their view on actions.
how to properly adopt AI so that it fits the strategic planning UC4 City Planning Office. The goal is to find discrepancies
process. We assume that adaptation of AI is inevitable. How- between a high-precision digital base map of the geogra-
ever, we observe that few sectors and industries understand AI phy and built environment of the municipality and actual
and its potential impact on organizations. aerial footage.
We have supported Jönköping Municipality (Sweden) in We aim to perform an intervention study in which we explore
developing an AI strategy. The city director and official man- the effects of conducting these types of AI-student projects as
agement are now implementing the AI strategy. We refer to this well as other bottom-up approaches, such as mentoring and
implementation as a top-down approach for AI transformation. competence development courses, in the public sector.

Authorized licensed use limited to: Norges Teknisk-Naturvitenskapelige Universitet. Downloaded on July 24,2021 at 16:07:25 UTC from IEEE Xplore. Restrictions apply.
 E. Ongoing Work in New Zealand [3] D. J. Teece, G. Pisano, and A. Shuen, “Dynamic capabilities and
strategic management,” Strategic Management Journal, vol. 18, no. 7,
We asked Waikato Regional Council to find environmental pp. 509–533, 1997.
use cases where AI can help to improve decision making. They [4] U. Gasser and V. A. Almeida, “A layered model for ai governance,”
provided four use cases, and now we have researchers from IEEE Internet Computing, vol. 21, no. 6, pp. 58–62, 2017.
[5] B. W. Wirtz, J. C. Weyerer, and C. Geyer, “Artificial intelligence and
the TAIAO project working on these cases. The Time-Evolving the public sector – applications and challenges,” International Journal
Data Science / Artificial Intelligence for Advanced Open Envi- of Public Administration, vol. 42, no. 7, pp. 596–615, 2019.
ronmental Science (TAIAO) is an AI program over seven years, [6] J. A. Schumpeter, The Theory of Economic Development. Harvard,
1934.
funded by the New Zealand Ministry of Business, Innovation, [7] R. Greenwood and C. R. Hinings, “Understanding radical organiza-
and Employment (MBIE). It is a collaboration between the tional change: Bringing together the old and the new institutionalism,”
Universities of Waikato, Auckland and Canterbury, Beca and Academy of Management Review, vol. 21, no. 4, pp. 1022–1054, 1996.
[8] D. Anderson and L. A. Anderson, Beyond change management: Ad-
MetService. vanced strategies for today’s transformational leaders. John Wiley &
Sons, 2002.
F. Ongoing Work in Norway [9] M. L. Tushman, W. H. Newman, and E. Romanelli, “Convergence and
We have been collaborating with several municipalities upheaval: Managing the unsteady pace of organizational evolution,”
California Management Review, vol. 29, no. 1, pp. 29–44, 1986.
within Norway on developing their AI capabilities, and ac- [10] A. Pettigrew, The Awakening Giant. Oxford: Blackwell, 1985.
cumulating the necessary resources to commence their AI [11] C. J. Gersick, “Revolutionary change theories: A multilevel exploration
transformation. An early study revealed some of the main of the punctuated equilibrium paradigm,” Academy of Management
Review, vol. 16, no. 1, pp. 10–36, 1991.
challenges municipalities face in their quest of deploying [12] H. Arazmjoo and H. Rahmanseresht, “A multi-dimensional meta-
AI applications [23]. The current work examines the role heuristic model for managing organizational change,” Management
of technological, organizational, and environmental enablers Decision, vol. 58, no. 3, pp. 526–543, 2019.
[13] H. Mintzberg and F. Westley, “Cycles of organizational change,” Strate-
and inhibitors of AI deployment in municipalities, as well as gic Management Journal, vol. 13, no. S2, pp. 39–59, 1992.
the realized organizational value from such investments. This [14] H. Tsoukas and R. Chia, “On organizational becoming: Rethinking
work, as well as ongoing research examining a process view organizational change,” Organization Science, vol. 13, no. 5, pp. 567–
582, 2002.
of AI deployment, aims to understand how the value of novel [15] P. Mikalef, R. van de Wetering, and J. Krogstie, “Building dynamic
technologies such as AI can be maximized. capabilities by leveraging big data analytics: The role of organizational
inertia,” Information & Management, 2020.
V. J OINT R ESEARCH AGENDA [16] P. Besson and F. Rowe, “Strategizing information systems-enabled
organizational transformation: A transdisciplinary review and new di-
We are aiming for an innovative research approach. We view rections,” Journal of Strategic Information Systems, vol. 21, no. 2, pp.
AI transformation as an interdisciplinary phenomenon, and we 103–124, 2012.
[17] E. Peretz-Andersson, N. Lavesson, and R. Torkar, “Ai transformation:
aim to investigate how organizations change due to AI technol- A systematic literature review,” 2021, submitted for peer-review.
ogy maturation and implementation. In this particular research [18] M. L. Ashwell, “The digital transformation of intelligence analysis,”
agenda, our main interest is in organizations which operate Journal of Financial Crime, vol. 24, no. 3, pp. 393–411, 2017.
[19] L. Heilig, E. Lalla-Ruiz, and S. Vob, “Digital transformation in maritime
in public sector. We aim to collaborate in an international ports: Analysis and a game theoretic framework,” Netnomics: Economic
setting, and to perform joint studies and comparative studies Research and Electronic Networking, vol. 18, no. 2–3, pp. 227–254,
across various contexts, as well as different public sector 2017.
[20] S. Akter, K. Michael, M. R. Uddin, G. McCarthy, and M. Rahman,
organizations and countries, including: Sweden, Norway, New “Transforming business using digital innovations: The application of ai,
Zealand, Germany, and The Netherlands. We believe that the blockchain, cloud and data analytics,” Annals of Operations Research,
interdisciplinary and international collaboration will yield a pp. 1–33, 2020.
[21] S. L. Wamba-Taguimdje, S. F. Wamba, J. R. K. Kamdjoug, and C. E. T.
comprehensive understanding of AI transformation and its Wanko, “Influence of artificial intelligence (ai) on firm performance: The
impact on the public sector. Also, we intend to establish a business value of ai-based transformation projects,” Business Process
foundation for decision and policy makers, so that they are Management Journal, 2020.
[22] B. Kitchenham, “Procedures for performing systematic reviews,” Keele
able to engage fruitfully in AI transformation. University. Technical Report TR/SE-0401, Department of Computer
Science, Keele University, UK, Tech. Rep., 2004.
ACKNOWLEDGMENT [23] P. Mikalef, S. O. Fjørtoft, and H. Y. Torvatn, “Artificial intelligence in
We are thankful to Jönköping Municipality and city director the public sector: A study of challenges and opportunities for norwegian
municipalities,” in Conference on e-Business, e-Services and e-Society,
Johan Fritz for opening up the organization to us and for 2019, pp. 267–277.
enabling us to interview decision makers, perform collabo-
rative pilot projects, and follow the strategic implementation
of digitalization and AI carried out by the municipality.
R EFERENCES
[1] D. Sculley, G. Holt, D. Golovin, E. Davydov, T. Phillips, D. Ebner,
V. Chaudhary, M. Young, J.-F. Crespo, and D. Dennison, “Hidden
technical debt in machine learning systems,” in Neural Information
Processing Systems, 2015.
[2] J. E. Dutton and S. E. Jackson, “Categorizing strategic issues: Links to
organizational action,” Academy of Management Review, vol. 12, no. 1,
pp. 76–90, 1987.

Authorized licensed use limited to: Norges Teknisk-Naturvitenskapelige Universitet. Downloaded on July 24,2021 at 16:07:25 UTC from IEEE Xplore. Restrictions apply.
View publication stats