Content uploaded by Matti Minkkinen
Author content
All content in this area was uploaded by Matti Minkkinen on Aug 17, 2023
Content may be subject to copyright.
Institutional logics underpinning AI governance
Fourteenth Scandinavian Conference on Information Systems (SCIS2023), Porvoo, Finland 1
THE INSTITUTIONAL LOGICS UNDERPINNING
ORGANIZATIONAL AI GOVERNANCE PRACTICES
Research paper
Minkkinen, Matti, University of Turku, Turku, Finland, matti.minkkinen@utu.fi
Mäntymäki, Matti, University of Turku, Turku, Finland, matti.mantymaki@utu.fi
Abstract
Recent developments in artificial intelligence (AI) promise significant benefits but also invoke novel
risks and harms to individuals, organizations, and societies. The rising role of AI necessitates effective
AI governance. However, translating AI ethics principles into governance practices remains challeng-
ing. Our paper recasts the “AI ethics translation problem” from a unidirectional translation process to
a bidirectional interaction between multiple institutional logics and organizational AI governance prac-
tices. We conduct a theory adaptation study using the AI governance translation problem as a domain
theory and institutional logics and institutional pluralism as method theories. Using this framework, we
synthesize key AI governance practices from the literature and outline four central institutional logics:
AI ethics principlism, managerial rationalism, IT professionalism, and regulatory oversight. The insti-
tutional logics and AI governance practices reciprocally influence one another: logics justify practices,
and practices enact logics. We provide an illustrative analysis of the ChatGPT chatbot to demonstrate
the framework. For future research, our conceptual study lays a framework for studying how plural
institutional logics drive AI governance practices and how practices can be used to negotiate conflicting
and complementary institutional logics.
Keywords: AI, AI governance, Institutional logics, IT governance.
1 Introduction
Recent developments in artificial intelligence (AI), fueled by growing quantities of data and increasingly
sophisticated processing algorithms, promise efficiency benefits across different sectors but also invoke
novel risks and potential harms to individuals and societies (Butcher & Beridze, 2019; Jobin et al., 2019;
Mäntymäki et al., 2022b). AI can be defined as the “frontier of computational advancements that refer-
ences human intelligence in addressing ever more complex decision-making problems” (Berente et al.,
2021). PricewaterhouseCoopers anticipates that AI could contribute 15.7 trillion dollars to the global
economy by 2030 (cited in Strümke et al., 2022). At the same time, algorithmic systems may enable
opaque discriminatory practices against minority groups (Keymolen, 2023), employees may face a loss
of jobs due to increasing automation (Palladino, 2022), and there are concerns over safety incidents in
AI systems (Falco et al., 2021). Due to the significant risks attached to AI technologies, AI regulation
is presently drafted in the European Union and elsewhere (Stix, 2022), and AI has become a matter of
organizational governance (Mäntymäki et al., 2022a; Schneider et al., 2022), distinct from previous IT
governance approaches (Brown & Grant, 2005; Tiwana et al., 2013).
Governance of AI, defined as the rules, practices, processes, and tools to ensure alignment of AI systems
and external requirements (Mäntymäki et al., 2022a), exhibits a paradox similar to the well-known pri-
vacy paradox, where individuals hold privacy in high esteem but do little to protect it in practice (Barth
& de Jong, 2017). AI governance seems to be simultaneously a crucial endeavor and relatively unim-
portant in the day-to-day operations of organizations. While the necessity of AI governance is acknowl-
edged to ensure the appropriate functioning and ethical and regulatory safeguards for AI systems,
Institutional logics underpinning AI governance
Fourteenth Scandinavian Conference on Information Systems (SCIS2023), Porvoo, Finland 2
empirical research on organizational AI governance indicates that organizations still devote little atten-
tion specifically to AI governance (Ibáñez & Olmeda, 2022; Papagiannidis et al., 2023; Stahl et al.,
2022).
In current research, this problem area is conceptualized as the “translation problem” or “principles-to-
practices gap” in AI ethics (Mittelstadt, 2019; Morley et al., 2020; Schiff et al., 2021). This refers to the
phenomenon where there are numerous sets of AI ethics principles (Jobin et al., 2019), but deducing
“concrete technological implementations from the very abstract ethical values and principles” remains
a major challenge (Hagendorff, 2020). While there are tools for ethical AI, most require more work to
be production-ready (Morley et al., 2020), and there are few guarantees that they cover the scope of the
ethics principles such as fairness and explainability. Moreover, given the multi-actor nature of AI gov-
ernance activities (Minkkinen et al., 2023), the translation problem of AI ethics indicates contradictions
that reach beyond organizations into their surrounding institutional forces (Alford & Friedland, 1991).
From these starting points, this study investigates the following research question:
How can the relationship between organizations’ AI governance requirements and AI governance prac-
tices be conceptualized?
We contribute to three scholarly domains. First, to the information systems (IS) literature (Brown &
Grant, 2005; Berente et al., 2021; Ågerfalk, 2020), we highlight AI governance that addresses the char-
acteristics of AI artifacts and their ethical risks. Second, we contribute to the literature on operational-
izing AI ethics in practice (Eitel-Porter, 2021; Ibáñez & Olmeda, 2022; Mittelstadt, 2019; Morley et al.,
2020; Seppälä et al., 2021; Stahl et al., 2022) by recasting the problem from a one-way “translation”
issue to a bidirectional dynamic between multiple institutional logics and AI governance practices,
where logics and practices reciprocally inform each other. Third, we contribute to the literature on multi-
actor AI governance (Butcher & Beridze, 2019; Clarke, 2019b; Gasser & Almeida, 2017; Kaminski &
Malgieri, 2021; Minkkinen et al., 2023; Shneiderman, 2020) by clarifying the institutional logics stem-
ming from the institutional environment and faced by organizations.
The remainder of the paper proceeds as follows. First, we establish the methodological and theoretical
background, positioning the study as a theory adaptation (Jaakkola, 2020) and discussing the domain
theory (AI ethics translation) and method theory (institutional logics and institutional pluralism). Then,
we present the practices and institutional logics in AI governance. We close with discussing how the
institutional logics and AI governance practices inform each other, offering an illustrative analysis of
the ChatGPT chatbot and articulating the implications of our study and future research directions.
2 Background
2.1 Theory adaptation
This article presents a theory adaptation (Jaakkola, 2020) study to revise the predominant understanding
of translating AI ethics into AI governance, using AI ethics translation as the domain theory and insti-
tutional logics and institutional pluralism as the method theory (Lukka & Vinnari, 2014). We draw on
two theoretical streams informing the theory adaptation: AI governance literature (e.g., Morley et al.,
2020; Schiff et al., 2021) and the literature on institutional logics and institutional pluralism (e.g., Ajer
et al., 2021; Alford & Friedland, 1991; Kraatz & Block, 2017). Following the principles of theory ad-
aptation (Jaakkola, 2020), we seek to change how AI ethics translation (domain theory) is viewed by
using institutional logics and institutional pluralism (method theory) (Lukka & Vinnari, 2014).
Within the institutional logic literature, we focus specifically on the IS literature on institutional logics
(Ajer et al., 2021; Berente & Yoo, 2012; Bernardi & Exworthy, 2020; Boonstra et al., 2018; Hansen &
Baroody, 2020), because institutional logics research has mushroomed in the management field and
because the IS literature is pertinent to AI governance where the governed IT artifact is central.
Institutional logics underpinning AI governance
Fourteenth Scandinavian Conference on Information Systems (SCIS2023), Porvoo, Finland 3
2.2 Domain theory: Translating AI ethics into practice through AI govern-
ance practices
AI ethics is predominantly approached through establishing and discussing guideline documents that
outline sets of principles (for overviews, see Thiebes et al., 2021; Hagendorff, 2020; Jobin et al., 2019).
Commonly referred principles include fairness, transparency, accountability, non-maleficence, and pri-
vacy (Jobin et al., 2019; Dignum, 2020). These principles deal with requirements for AI systems, for
example, that they should not discriminate against ethnic groups or genders (fairness) or that the opera-
tions of AI systems should be sufficiently visible to users and experts (transparency). Due to the abstract
nature of the AI ethics principles, the translation of AI ethics principles into AI development and use
practices has been the topic of recent scholarly attention (Mittelstadt, 2019; Morley et al., 2020; Li et
al., 2023; Morley et al., 2023; Schiff et al., 2021). The AI ethics translation problem refers to how ab-
stract ethical principles, such as fairness, could ensure responsible design, development, and use of AI
systems. This entails translating AI ethics to AI governance, understood as the structures, processes, and
tools that enable responsible design and use of AI (Mäntymäki et al., 2022a).
Researchers have noted that translating AI ethics principles into AI governance practices has been prob-
lematic. Even though software tools exist, few comprehensive solutions are available for organizational
use (e.g., Morley et al., 2020; Schiff et al., 2021). Some scholars criticize the principle-based perspective
for its high abstraction level and inability to account for the diverse application domains of AI systems
(Mittelstadt, 2019; Morley et al., 2023). Others note that principles could nevertheless guide organiza-
tions’ business processes (Clarke, 2019a) and the professional norms of AI developers (Seger, 2022).
IT governance research (Brown & Grant, 2005; Tiwana et al., 2013; Weill & Ross, 2004) provides one
entry point into translating AI ethics into AI governance. IT governance research within the IS field has
looked into, for example, IT governance’s dimensions (Tiwana et al., 2013), antecedents and conse-
quences (Bradley et al., 2012), business/IT alignment (De Haes & Van Grembergen, 2009), governance
archetypes (Weill & Ross, 2004), and contingency factors (Brown & Grant, 2005). In a nutshell, IT
governance is about specifying decision rights and accountabilities and ensuring desirable IT use and
regulatory compliance (Brown & Grant, 2005; Weill & Ross, 2004). These aspects are also pertinent to
AI governance because AI systems are IT systems with particular characteristics (Berente et al., 2021;
Mäntymäki et al., 2022a).
However, current AI technologies exhibit features that challenge IT governance frameworks, necessi-
tating AI governance as a separate concern. As incidents of biased and unsafe AI show (Wei & Zhou,
2023), ethical concerns and risks of misuse are more severe in the case of AI. For example, biases against
ethnic minorities in facial recognition technologies have been widely critiqued (Raji et al., 2020). The
amplified risks stem from the nature of AI systems as IT artifacts. AI systems act increasingly inde-
pendently from human oversight, they improve by learning from data, and their workings are inscrutable
to the wider public and often even to their developers (Berente et al., 2021). Hence, the increasingly
agentic nature of AI systems as artifacts (Baird & Maruping, 2021; Ågerfalk, 2020) warrants research
attention beyond adapting existing IT governance frameworks that govern less agentic IT artifacts. Be-
cause AI ethics principles (Jobin et al., 2019; Dignum, 2020) tackle AI-specific issues, their translation
into governance remains a valid starting point for AI governance despite its challenges.
What is relevant to the present paper is that the translation perspective to AI ethics makes certain key
assumptions about the domain of responsible AI. First, the principle-based approach requires a near-
consensus on high-level principles. This is achieved relatively well at present, even though there are
several sets of AI ethics guidelines (Jobin et al., 2019). Nonetheless, tensions between different princi-
ples or between principles and other requirements have received little attention (Whittlestone et al.,
2019). Second, the translation perspective assumes that the role of ethical principles is to serve as a
starting point for translation into practical governance, which is only one possible function of ethical
principles that can also inform professional culture more indirectly (Seger, 2022). Third, the translation
perspective provides little insight into the nature, foundations, and operating mechanisms of ethics
Institutional logics underpinning AI governance
Fourteenth Scandinavian Conference on Information Systems (SCIS2023), Porvoo, Finland 4
principles and possible competing considerations. In other words, AI ethics translation lacks a clear
theoretical foundation as a phenomenon for IS research and other socio-technical scholarship.
While the debate on translating AI ethics is ongoing, it is clear that new approaches could enable re-
searchers to elaborate particularly on tensions between ethical principles and other relevant considera-
tions. Therefore, this paper presents institutional logics and institutional pluralism as method-theoretical
lenses (Jaakkola, 2020; Lukka & Vinnari, 2014) through which the AI ethics translation problem can be
revisited and made more theoretically understandable and practically tractable.
2.3 Method theory: Institutional logics and institutional pluralism
The institutional perspective on organizations and IS starts from the premise that organizations are con-
strained by social structures and forces in the form of several institutions rather than a monolithic “so-
ciety” (Alford & Friedland, 1991). From an institutional theory perspective, institutional pressure drives
organizations to implement AI governance practices. Institutional pressure (Figure 1) can be divided
into strong institutional pressure, which refers to involuntary compliance with significant punitive con-
sequences from nonconformity, and weak institutional pressure, where adoption is voluntary and non-
conformity does not carry significant consequences (Berente et al., 2019).
Figure 1. Institutional pressures for AI governance
Institutional logics are collective belief systems that shape actors’ cognition and behavior. They are
socially constructed sets of principles, practices, beliefs, rules, and systems through which organizations
and individuals make sense of their social reality and of appropriate behaviors (Greenwood et al., 2011;
Thornton et al., 2012). Institutional logics are described with varying terms as “broader cultural tem-
plates” (Pache & Santos, 2010), “socially constructed patterns” leveraged during action-taking (Ajer et
al., 2021), and “socially constructed sets of practices, beliefs, rules and systems” (Thornton et al., 2012).
The emphasis on individual and collective cognition and the normative binding nature of institutional
logics are common features across these conceptualizations.
Institutional pluralism means a situation where organizations face institutional pressures from multiple
logics that may complement each other or be in conflict, i.e., incongruent (Ocasio et al., 2017). Berente
et al. (2019) conceptualize incongruent institutional logics as a set of logics that, utilized in a situation,
cannot guide an actor’s practices “without creating a dissonance that calls for fundamentally changing
those practices.” Incongruent institutional logics, thus, demand an active response from organizational
actors to avoid dissonance leading to paralysis. Plural institutional logics help explain seemingly con-
tradictory organizational behaviors because organizational actors “do not simply apply institutional
rules, but they navigate and engage institutional orders in their everyday practices in ways that are con-
sistent with particular logics” (Berente & Yoo, 2012). However, in recent work on institutional logics,
Institutional logics underpinning AI governance
Fourteenth Scandinavian Conference on Information Systems (SCIS2023), Porvoo, Finland 5
researchers emphasize that logics can also complement one another, creating alignment and organiza-
tional abilities to function across institutional domains (Hansen & Baroody, 2020).
The closely related concept of institutional complexity deals with organizational responses to conflicting
demands posed by different institutional logics (Greenwood et al., 2011; Pache & Santos, 2010). A
central divide in the literature is between focusing on organizational strategies to respond to institutional
pluralism and organizational structures and practices in response to pluralism (Greenwood et al., 2011;
Kraatz & Block, 2017). In this paper, we focus on the latter, practice-based approach to organizations’
tackling of institutional pluralism and complexity.
Within the IS field, institutional logics have been used to study the implementation of enterprise systems
(Berente et al., 2019), electronic health records (Hansen & Baroody, 2020), and enterprise architecture
(Ajer et al., 2021). IS studies on institutional logics tend to fall under three basic types. First, studies
have been conducted on the intra-organizational implementation of enterprise systems and enterprise
architectures, where a new system is introduced into complex organizational systems with pre-existing
institutional logics (Ajer et al., 2021; Berente & Yoo, 2012; Berente et al., 2019). In this case, the po-
tential conflict between logics is between the new IS’s logic and the context-specific existing logics, and
the role of IT artifacts is left somewhat implicit. Second, IS researchers have used the institutional logics
lens to study how logics, IT affordances, and organizational attention are intertwined in institutional
change, particularly in times of crisis (Faik et al., 2020; Oborn et al., 2021). In this set of studies, the
institutional logics are broader, referring to logics like the state logic or the family logic, and the role of
the artifact is theorized more extensively than in the first set of studies. Third, IS research has examined
multi-stakeholder situations in IT innovation and system adoption (Bernardi & Exworthy, 2020;
Boonstra et al., 2018; Hansen & Baroody, 2020). In this case, different stakeholder groups are viewed
as proponents of different logics, while the logics are context-specific, as in the first group of studies.
Our study aligns most closely with the third set of studies because AI governance is usually not a specific
system that is implemented, and the logics involved are context-specific rather than broad societal logics,
such as the family logic. However, because we discuss AI governance across different sectors, we
bracket out any sector-specific institutional logics, such as healthcare professionalism, from this paper’s
investigation. IS studies on institutional logics have thus far focused on the healthcare sector (Ajer et
al., 2021; Bernardi & Exworthy, 2020; Boonstra et al., 2018; Hansen & Baroody, 2020; Oborn et al.,
2021), apart from a few notable exceptions (Berente & Yoo, 2012; Berente et al., 2019). Therefore,
sector-specific institutional logics, such as healthcare professionalism, have been central in considering
institutional pluralism, unlike our case, which discusses AI governance across sectors.
3 Results
3.1 AI governance practices
The operationalization of ethical AI occurs in organizations that deploy and develop AI systems. There-
fore, it is important to focus on AI governance at the level of organizational practices (Mäntymäki et al.,
2022a) which give meaning and thematic coherence to activities that may appear trivial by themselves
(Smets et al., 2012). Table 1 outlines the organizational AI governance practices in the literature.
Table 1. Organizational AI governance practices
Practice
Sources
AI auditing
Mökander et al., 2021; Minkkinen et al., 2022a
Competence and knowledge development
Seppälä et al., 2021
Corporate sustainability reporting
Minkkinen et al., 2022b; Sætra, 2023
Institutional logics underpinning AI governance
Fourteenth Scandinavian Conference on Information Systems (SCIS2023), Porvoo, Finland 6
Data governance and data management
Schneider et al., 2022; Seppälä et al., 2021; Stahl et al.,
2022
Explainability and transparency practices
Brundage et al., 2020; Laato et al., 2022b; Meske et al.,
2022
Impact assessment
Kaminski & Malgieri, 2021; Metcalf et al., 2021
Organizational policies and ethics guidelines
Schneider et al., 2022; Seppälä et al., 2021
Regulatory compliance
Schneider et al., 2022
Risk management
Stahl et al., 2022; Tournas & Bowman, 2021
Software engineering workflows, AI design
and development
Laato et al., 2022a; Shneiderman, 2020; Seppälä et al.,
2021; Stahl et al., 2022
Stakeholder collaboration
Schneider et al., 2022; Seppälä et al., 2021; Zhu et al.,
2021
Standardization
Cihon, 2019; Stahl et al., 2022
Validation, testing, and verification
Brundage et al., 2020; Yeung et al., 2020; Zhu et al., 2021
The literature shows that AI governance practices range from software engineering to regulatory exper-
tise. Some practices, such as explainability and transparency, are related to one ethical principle, but
generally, there is no clear relationship between AI ethics principles and governance practices.
3.2 Institutional logics in AI governance
In Table 2, we present four institutional logics in AI governance and their dimensions. Institutional
logics are usually divided into dimensions. In the IS literature, often mentioned dimensions include
principles (Ajer et al., 2021; Berente & Yoo, 2012; Berente et al., 2019), assumptions (Ajer et al., 2021;
Berente & Yoo, 2012; Berente et al., 2019; Hansen & Baroody, 2020), identity (Berente & Yoo, 2012;
Berente et al., 2019; Bernardi & Exworthy, 2020; Boonstra et al., 2018; Hansen & Baroody, 2020), and
domain (Berente & Yoo, 2012; Berente et al., 2019; Hansen & Baroody, 2020), drawing primarily on
Thornton and Ocasio (2008). Other dimensions include sources of legitimacy (Ajer et al., 2021; Bernardi
& Exworthy, 2020; Boonstra et al., 2018; Hansen & Baroody, 2020), sources of authority (Bernardi &
Exworthy, 2020; Boonstra et al., 2018; Hansen & Baroody, 2020), and basis of attention (Bernardi &
Exworthy, 2020; Boonstra et al., 2018). In this paper, for the sake of parsimony, we adopt four com-
monly used dimensions: principles, assumptions, identity, and sources of legitimacy.
Table 2. Four institutional logics in AI governance
Dimensions
Logic of AI ethics
principlism
Logic of managerial
rationalism
Logic of IT pro-
fessionalism
Logic of regula-
tory oversight
Principles
Adherence to high-
level principles and
guidelines (fairness,
non-maleficence, ac-
countability, and pri-
vacy) (Jobin et al.,
2019; Seger, 2022).
Embedding principles
in practices of AI de-
velopment and
Accountability and
control (Berente &
Yoo, 2012).
Competition, effi-
ciency, cost control,
continuous improve-
ment (Hansen &
Baroody, 2020).
Data usability, data
management, tech-
nical effectiveness,
user participation
(Hansen &
Baroody, 2020).
Compliance, cost
control, efficiency,
standardization,
continuous im-
provement (Hansen
& Baroody, 2020)
Institutional logics underpinning AI governance
Fourteenth Scandinavian Conference on Information Systems (SCIS2023), Porvoo, Finland 7
professional culture
(Mittelstadt, 2019;
Seger, 2022).
Assumptions
Normative consensus
provided by abstract
principles (Gasser &
Almeida, 2017; Res-
seguier & Rodrigues,
2021).
Accountability and
control through stand-
ardization and visibil-
ity (Berente & Yoo,
2012).
Financial incentives
drive behavior (Han-
sen & Baroody,
2020).
Benchmarking sup-
ports the improve-
ment of outcomes
(Hansen & Baroody,
2020).
Information inte-
gration improves
outcomes, IT for
IT’s sake will fail
(Hansen &
Baroody, 2020).
Incentives and pen-
alties guide behav-
ior (Hansen &
Baroody, 2020).
Identity
Ethics experts discern
the relevant issues and
provide a foundation
for operationalization
(cf. Morley et al.,
2020).
Standardized struc-
ture implies rational
bureaucracy and ob-
jective criteria for re-
source allocation
(Berente & Yoo,
2012).
IT offers transpar-
ency and precision
in a standardized
way (Boonstra et
al., 2018).
Information bro-
kers, vendors (Han-
sen & Baroody,
2020)
Rules empower
regulatory actors
such as regulators
and administrators
(Hansen &
Baroody, 2020).
Sources of
legitimacy
Expertise in AI tech-
nologies and AI ethics
issues (Palladino,
2022), ethics guide-
lines (Jobin et al.,
2019).
Managerial roles, per-
formance manage-
ment (Bernardi & Ex-
worthy, 2020).
Data evidence, objec-
tive data (Hansen &
Baroody, 2020).
Financial outcomes,
profitability, and sur-
vival (Hansen &
Baroody, 2020).
Education, rational
standards based on
a technical
worldview, sys-
tems sciences
(Boonstra et al.,
2018).
Binding legislative
documents.
Standardization:
establishing gener-
ally accepted
standards (Hansen
& Baroody, 2020)
The dimensions of the AI governance institutional logics are compiled from previous literature on com-
parable fields. While institutional logics have been discussed in IT governance (Boonstra et al., 2018;
Offenbeek et al., 2013), work on institutional logics within AI governance is incipient. In a review of
AI governance principles and their operationalization in practical tools, the institutional logic of private
companies and the “tech community” were seen to inscribe a narrow technology- and solution-oriented
institutional logic to the implementation of AI ethics (Palladino, 2022). Moreover, technical tools and
social governance arrangements are deemed to be poorly integrated at present (Palladino, 2022).
AI ethics principlism. AI ethics principlism can be viewed as an institutional logic because one of its
primary functions is to influence mindsets, professional culture, and professional norms in AI design
and development (Seger, 2022). Principlism is closely related to the deontological approach to applied
ethics, that is, adherence to rules and duties (Hagendorff, 2020). There is an abundance of research,
particularly conceptual research, on the so-called translation problem or principles-to-practices gap in
AI governance (Morley et al., 2020; Palladino, 2022; Schiff et al., 2021). Nevertheless, to the best of
our knowledge, AI ethics principlism has not been previously conceptualized as an institutional logic.
Institutional logics underpinning AI governance
Fourteenth Scandinavian Conference on Information Systems (SCIS2023), Porvoo, Finland 8
Principlism refers to the practice of using ethical principles in tackling moral problems that arise in real-
world situations, and it has been particularly discussed in medicine and bioethics (Clouser & Gert, 1990).
Principlism in medical ethics embeds core norms into professional practice and helps to identify ethical
challenges, guide health policy, and support clinical decision-making (Mittelstadt, 2019). AI ethics prin-
ciplism refers to a focus on adherence to core principles and ethical guidelines as the preferred approach
to dealing with AI ethics problems as well as the embedding of principles in professional practice in AI
development (Seger, 2022; Mittelstadt, 2019; cf. Jobin et al., 2019; Schiff et al., 2021).
A principled approach to AI ethics operates on the level of abstract principles by design and neglects
specific features of contexts and practices (Resseguier & Rodrigues, 2021). The list of core principles
is relatively established, consisting of transparency, fairness, non-maleficence, accountability/responsi-
bility, and privacy, sometimes coupled with less-often-mentioned principles such as human dignity
(Jobin et al., 2019; Palladino, 2022). However, a feature of AI ethics principlism institutional logic is
the abstract nature and lack of a clear definition of the principles (Palladino, 2022). This implies an
assumption that the search for a normative consensus (Gasser & Almeida, 2017) on abstract principles
is more important than clear definitions of the content and implications of principles.
The focus on ethical principles as guidance and legitimation means there is high functional indetermi-
nation, and roles and accountabilities in AI governance are typically unclear (Radu, 2021). In addition
to guidelines, AI ethics principlism primarily uses AI experts and AI ethics experts as a source of legit-
imacy (Palladino, 2022). In particular, there is a tendency to form consultative bodies, such as AI ethics
boards, with loosely defined mandates and opaque criteria for selecting relevant experts (Radu, 2021).
AI ethics principlism is motivated by promoting trust toward new technologies and, thereby, realizing
their full potential (Palladino, 2022). The counterpart is the fear of stifling innovation with overly strict
rules; hence, the focus is on guiding principles (Radu, 2021). While in the medical field, principlism has
a strong institutional backing in curricula and medical practice, as well as accountability mechanisms
such as ethics committees, AI ethics principlism is more loosely organized at present (Mittelstadt, 2019).
In the literature on the translation problem, AI ethics principlism is not discussed in “pure form” because
it is already implicitly mixed with other logics, such as managerialism and IT development (Morley et
al., 2020; Palladino, 2022). For example, the findings of Palladino (2022) can be interpreted as IT pro-
fessionalism and managerialism subverting AI ethics principles and turning social issues into techno-
logically solvable problems for legitimation (gaining trust) and self-interest (avoiding regulatory fines).
Managerial rationalism. The managerial rationalism logic is omnipresent in organizations, and it em-
phasizes accountability, control, standardization, and business-like efficiency (Berente & Yoo, 2012).
Managerialism also prescribes self-preservation through effective business management (Hansen &
Baroody, 2020). Logically, then, attention is turned to financial performance, cost control, revenue gen-
eration, and profitability (Hansen & Baroody, 2020). The legitimacy of this logic comes from objective
measures of efficiency and profitability.
IT professionalism. The rationality is focused on the instrumentality of IT (Boonstra et al., 2018), where
developers design systems that model reality and become useful tools for managers and others to achieve
common ends (Boonstra et al., 2018). The logic places emphasis on the role of technical knowledge and
on being in control, and it is concerned with using appropriate development methodologies, such as
iterative and agile (Boonstra et al., 2018), as well as the quality of technical solutions and information
system design (Hansen & Baroody, 2020). There is an overarching focus on achieving high-speed meth-
ods and a sufficient pace of development to meet increasing time pressures (O’Connor et al., 2023).
Regulatory oversight. The regulatory oversight logic is comparatively clear-cut because it refers to the
effects of more binding rules than the principles discussed under AI ethics principlism. It is thus more
coercive in nature than the other logics. Under the regulatory oversight logic, individuals or organiza-
tions identify compliance with existing and anticipated laws, governing bodies, and industry practices
(Hansen & Baroody, 2020). Informed by this logic, organizations take actions because they believe that
they have to and that they are mandated to act in a certain way (Hansen & Baroody, 2020). The regula-
tory oversight logic is evident in AI governance because new AI legislation is underway, most notably
the upcoming AI Act in the European Union (Minkkinen et al., 2023; Stix, 2022).
Institutional logics underpinning AI governance
Fourteenth Scandinavian Conference on Information Systems (SCIS2023), Porvoo, Finland 9
4 Discussion
4.1 Institutional logics and AI governance practices
Drawing together the themes from the previous sections, Figure 2 outlines the bidirectional interaction
between institutional logics and AI governance practices. Rather than a one-way translation from AI
ethics principles to governance practices, multiple institutional logics (including AI ethics principlism)
and governance practices continuously influence one another. Institutional logics justify and symboli-
cally ground governance practices, i.e., logics provide the organizing principle and rationality as well as
normative legitimation to the conduct of practices (Alford & Friedland, 1991; Berente & Yoo, 2012).
The logics imbue the practices with meaning, making it sensible to devote time and attention to actions
related to AI auditing, for example, even when not strictly required by legislation. In the other direction,
practices materially enact the socio-cognitive institutional logics by deriving the practical implications
from the logics, i.e., what activities should be conducted according to the logic (Boonstra et al., 2018;
Hansen & Baroody, 2020; Smets et al., 2012). The notion of enactment comes close to the translation
of AI ethics principles. However, in the case of institutional logics, AI ethics principlism is only one
logic, offset by other logics. Therefore, AI ethics translation is not a one-directional issue but rather a
bidirectional and dynamic interaction between institutional logics and AI governance practices.
Figure 2. Dynamic interaction between institutional logics and AI governance practices
Figure 2 presents an abstract model where institutional logics, as a single unit, influences a single unit
of AI governance practices. In reality, the relationship is not one-to-one but many-to-many. In other
words, institutional logics and AI governance practices can be seen to form a matrix with AI governance
practices as rows and institutional logics as columns. In an empirical study, the cells of this kind of
matrix could be filled with indications of either support (e.g., with “+” symbols) or undermining (e.g.,
with “-” symbols) for each pair of logic and practice. Then, the rows would show which governance
practices are strongly supported by logics (many “+” symbols), undermined by logics (many “-” sym-
bols), or ambiguously supported (both symbols present).
Our conceptual study does not provide evidence of how individual logics support or undermine govern-
ance practices. However, some preliminary hypotheses could be made. For instance, AI auditing and
impact assessment align with AI ethics principlism, and regulatory compliance aligns with regulatory
oversight. Standardization is likely to be supported by managerial rationalism, IT professionalism, and
regulatory oversight. On the other hand, auditing and impact assessment may conflict with managerial
rationalism because they demand time and resources – ethics costs (Mittelstadt, 2019).
The degree of incongruence between institutional logics can be seen as the driving force of AI govern-
ance practices because it provides the central tension that requires organizations to act (to avoid
conflicting with one or more logic) but makes action complex due to pluralism. Rather than a situation
of “old” logics and a “new” logic introduced by an implemented IS (Berente et al., 2019; Berente &
Institutional logics underpinning AI governance
Fourteenth Scandinavian Conference on Information Systems (SCIS2023), Porvoo, Finland 10
Yoo, 2012; Ajer et al., 2021), the AI governance case exhibits a “force field” of multiple institutional
logics that organizations negotiate by establishing and modifying their AI governance practices.
Practices, then, can mediate the contradictions and complementarities between institutional logics. One
way of reconciling conflicting logics is loose coupling. Practices may be loosely coupled with institu-
tional logics, which means that practices are not in conflict with logics, but they only loosely enact the
logics, for example, by ceremonially enacting certain principles (Berente & Yoo, 2012). Another way
of reconciling competing or complementary institutional logics is “reticulation,” that is, the intertwining
of practices (motivated by different institutional logics) via activities shared by both practices (Hansen
& Baroody, 2020). Reticulation means that a higher-level practice, such as risk management, can have
shared lower-level activities with another practice, such as AI auditing (Hansen & Baroody, 2020).
4.2 Illustrative analysis of ChatGPT
Even though our study is conceptual, we present an illustrative analysis of ChatGPT to test the applica-
bility of our findings. ChatGPT is an AI chatbot developed by OpenAI based on a large language model
(OpenAI, 2023a). It is able to ask questions and learn from human feedback, making its interaction with
users seem more human-like than previous chatbots (Chatterjee & Dethlefs, 2023). ChatGPT and similar
applications have been discussed under the term generative AI, meaning technologies that use deep
learning to generate human-like content in response to complex and varied prompts (Lim et al., 2023).
ChatGPT is a fruitful case because the force field of congruences and incongruences between institu-
tional logics is readily visible. From the AI ethics principlism perspective, already in 2018, OpenAI
published a charter (OpenAI, 2018) that outlines four principles: broadly distributed benefits, long-term
safety, technical leadership, and cooperative orientation. The company’s product safety standards
(OpenAI, 2023b), in turn, list the principles of minimizing harm, building trust, learning and iterating,
and being a pioneer in trust and safety. The translation problem is evident here: How to turn these prin-
ciples into AI governance practices?
For ChatGPT, the AI ethics principlism logic is supported by the regulatory oversight logic, where pol-
icy instruments such as the EU’s AI Act (Stix, 2022) reinforce the principles of minimizing harm and
dealing with safety. In addition, the IT professionalism logic supports AI ethics principlism because
both logics espouse standardized practices and clear rules. Thus, AI ethics principlism, regulatory over-
sight, and IT professionalism all support standardization and sensible risk management practices.
However, the managerial rationalism logic strongly drives ChatGPT as a product. Although OpenAI
was initially established as a non-profit, it quickly became a for-profit company. In January 2023, Mi-
crosoft invested $10 billion in OpenAI, and the company seeks to integrate ChatGPT into its products,
such as Microsoft Office (Spataro, 2023). Managerial rationalism is allied with IT professionalism in
seeking technical efficiency, which may facilitate effective AI governance practices or undermine them.
AI ethics principlism and regulatory oversight, which espouse time-intensive governance practices, con-
flict with managerial rationalism and IT professionalism, which favor speed and agility. Hence, the
temporal dimension of institutional logics is a central point of incongruence in the ChatGPT case. In-
deed, generative AI looks to be a new frontier in the so-called AI race where fast movers reap significant
benefits compared to companies with heavy governance practices.
In sum, ethical principles are in place for ChatGPT, but a crucial question is how regulatory oversight
and IT professionalism could support governance practices given the intense time pressure expressed
by managerial rationalism and IT professionalism. Reticulating activities across different practices (and
logics) (Hansen & Baroody, 2020) could decrease the risk of mere ceremonial adherence to principles.
4.3 Implications for theory and practice
In relation to IS literature on IT governance (Brown & Grant, 2005; Tiwana et al., 2013) and AI (Berente
et al., 2021; Ågerfalk, 2020), we call for an approach to governance that addresses the characteristics of
AI artifacts and the concomitant ethical risks and tensions. To the literature on operationalizing AI ethics
in practice (Eitel-Porter, 2021; Ibáñez & Olmeda, 2022; Mittelstadt, 2019; Morley et al., 2020; Seppälä
Institutional logics underpinning AI governance
Fourteenth Scandinavian Conference on Information Systems (SCIS2023), Porvoo, Finland 11
et al., 2021; Stahl et al., 2022), we recast the unidirectional translation problem into an issue of multiple
institutional logics and AI governance practices reciprocally influencing each other. While this makes
the issue theoretically complex, it is more true to the continuous justification of AI governance practices
and enactment of institutional logics than the AI ethics translation perspective. To the literature on multi-
actor AI governance (Butcher & Beridze, 2019; Clarke, 2019b; Gasser & Almeida, 2017; Kaminski &
Malgieri, 2021; Minkkinen et al., 2023; Shneiderman, 2020), we enumerate the core institutional logics
that drive AI governance. It is widely known that regulation and market pressures influence the govern-
ance of technologies. However, conceptualizing these pressures as institutional logics gives additional
theoretical tools to discuss institutional pluralism and organizational responses to it.
In terms of practical implications, the results underscore the conflicting demands involved in AI gov-
ernance from a management perspective. Instead of translating AI ethics principles as non-functional
requirements into AI systems and their use, planning and conducting AI governance practices requires
the continuous balancing of conflicting and complementary logics.
4.4 Limitations and future research directions
As a conceptual theory adaptation study (Jaakkola, 2020), this study can present preliminary findings to
be tested in subsequent empirical studies and further theoretical elaboration. First, cross-sectoral, eth-
nographic, and longitudinal studies of organizational AI governance are warranted to validate the find-
ings and dig deeper into the role of institutional logics in AI governance practices. Second, future studies
could investigate the organizational responses and strategies available to organizations for tackling in-
stitutional pluralism (Greenwood et al., 2011; Pache & Santos, 2010). Third, this paper's lists of
institutional logics and AI governance practices are not exhaustive. Other logics and practices could be
discovered through literature reviews and empirical studies. However, in the interest of theoretical par-
simony, as few significant logics as possible should be considered because the level of complexity rises
as concepts are added. Thus, adding institutional logics may not yield new insights into how institutional
logics and AI governance practices condition one another. Fourth and finally, the appropriate level of
theoretical abstraction in studying AI governance remains for future studies to address.
References
Ågerfalk, P. J. (2020). Artificial intelligence as digital agency. European Journal of Information Sys-
tems, 29(1), 1-8. https://doi.org/10.1080/0960085X.2020.1721947
Ajer, A. K. S., Hustad, E., & Vassilakopoulou, P. (2021). Enterprise architecture operationalization and
institutional pluralism: The case of the Norwegian Hospital sector. Information Systems Journal,
31(4), 610-645. https://doi.org/10.1111/isj.12324
Alford, R. R., & Friedland, R. (1991). Bringing society back in: Symbols, practices, and institutional
contradictions. In W. W. Powell & P. J. DiMaggio (Eds.), The new institutionalism in organisational
analysis (pp. 232-267). University of Chicago Press.
Baird, A., & Maruping, L. M. (2021). The Next Generation of Research on IS Use: A Theoretical Frame-
work of Delegation to and from Agentic IS Artifacts. MIS Quarterly, 45(1), 315-341.
https://doi.org/10.25300/misq/2021/15882
Barth, S., & de Jong, M. D. T. (2017). The privacy paradox – Investigating discrepancies between ex-
pressed privacy concerns and actual online behavior – A systematic literature review. Telematics and
Informatics, 34(7), 1038-1058. https://doi.org/10.1016/j.tele.2017.04.013
Berente, N., Gu, B., Recker, J., & Santhanam, R. (2021). Managing artificial intelligence. MIS Quar-
terly, 45(3), 1433-1450. https://doi.org/10.25300/MISQ/2021/16274
Berente, N., Lyytinen, K., Yoo, Y., & Maurer, C. (2019). Institutional logics and pluralistic responses
to enterprise system implementation: a qualitative meta-analysis. MIS Quarterly, 43(3), 873-902.
https://doi.org/10.25300/MISQ/2019/14214
Institutional logics underpinning AI governance
Fourteenth Scandinavian Conference on Information Systems (SCIS2023), Porvoo, Finland 12
Berente, N., & Yoo, Y. (2012). Institutional Contradictions and Loose Coupling: Postimplementation
of NASA’s Enterprise Information System. Information Systems Research, 23(2), 376-396.
https://doi.org/10.1287/isre.1110.0373
Bernardi, R., & Exworthy, M. (2020). Clinical managers’ identity at the crossroad of multiple institu-
tional logics in it innovation: The case study of a health care organization in England. Information
Systems Journal, 30(3), 566-595. https://doi.org/10.1111/isj.12267
Boonstra, A., Yeliz Eseryel, U., & van Offenbeek, M. A. G. (2018). Stakeholders’ enactment of com-
peting logics in IT governance: polarization, compromise or synthesis. European Journal of Infor-
mation Systems, 27(4), 415-433. https://doi.org/10.1057/s41303-017-0055-0
Bradley, R. V., Byrd, T. A., Pridmore, J. L., Thrasher, E., Pratt, R. M. E., & Mbarika, V. W. A. (2012).
An Empirical Examination of Antecedents and Consequences of IT Governance in US Hospitals.
Journal of Information Technology, 27(2), 156-177. https://doi.org/10.1057/jit.2012.3
Brown, A. E., & Grant, G. G. (2005). Framing the Frameworks: A Review of IT Governance Research.
Communications of the Association for Information Systems, 15.
https://doi.org/10.17705/1cais.01538
Brundage, M., Avin, S., Wang, J., Belfield, H., Krueger, G., Hadfield, G., Khlaaf, H., Yang, J., Toner,
H., Fong, R., Maharaj, T., Koh, P. W., Hooker, S., Leung, J., Trask, A., Bluemke, E., Lebensold, J.,
O’Keefe, C., Koren, M., . . . Anderljung, M. (2020). Toward Trustworthy AI Development: Mecha-
nisms for Supporting Verifiable Claims. arXiv. http://arxiv.org/abs/2004.07213
Butcher, J., & Beridze, I. (2019). What is the State of Artificial Intelligence Governance Globally. The
RUSI Journal, 164(5-6), 88-96. https://doi.org/10.1080/03071847.2019.1694260
Chatterjee, J., & Dethlefs, N. (2023). This new conversational AI model can be your friend, philosopher,
and guide. and even your worst enemy. Patterns (N Y), 4(1), 100676. https://doi.org/10.1016/j.pat-
ter.2022.100676
Cihon, P. (2019). Standards for AI Governance: International Standards to Enable Global Coordination
in AI Research & Development.
Clarke, R. (2019a). Principles and business processes for responsible AI. Computer Law & Security
Review, 35(4), 410-422. https://doi.org/10.1016/j.clsr.2019.04.007
Clarke, R. (2019b). Regulatory alternatives for AI. Computer Law & Security Review, 35(4), 398-409.
https://doi.org/10.1016/j.clsr.2019.04.008
Clouser, K. D., & Gert, B. (1990). A critique of principlism. J Med Philos, 15(2), 219-236.
https://doi.org/10.1093/jmp/15.2.219
De Haes, S., & Van Grembergen, W. (2009). An Exploratory Study into IT Governance Implementa-
tions and its Impact on Business/IT Alignment. Information Systems Management, 26(2), 123-137.
https://doi.org/10.1080/10580530902794786
Dignum, V. (2020). Responsibility and artificial intelligence. In M. D. Dubber, F. Pasquale, & S. Das
(Eds.), The Oxford handbook of ethics of AI (pp. 213-231). Oxford University Press. https://ox-
fordhandbooks.com/view/10.1093/oxfordhb/9780190067397.001.0001/oxfordhb-9780190067397-
e-12
Eitel-Porter, R. (2021). Beyond the promise: implementing ethical AI. AI and Ethics, 1(1), 73-80.
https://doi.org/10.1007/s43681-020-00011-6
Faik, I., Barrett, M., & Oborn, E. (2020). How Information Technology Matters in Societal Change: An
Affordance-Based Institutional Logics Perspective. MIS Quarterly, 44(3), 1359-1390.
https://doi.org/10.25300/MISQ/2020/14193
Falco, G., Shneiderman, B., Badger, J., Carrier, R., Dahbura, A., Danks, D., Eling, M., Goodloe, A.,
Gupta, J., Hart, C., Jirotka, M., Johnson, H., LaPointe, C., Llorens, A. J., Mackworth, A. K., Maple,
C., Pálsson, S. E., Pasquale, F., Winfield, A., . . . Yeong, Z. K. (2021). Governing AI safety through
independent audits. Nature Machine Intelligence, 3(7), 566-571. https://doi.org/10.1038/s42256-
021-00370-7
Gasser, U., & Almeida, V. A. F. (2017). A Layered Model for AI Governance. IEEE Internet Compu-
ting, 21(6), 58-62. https://doi.org/10.1109/mic.2017.4180835
Institutional logics underpinning AI governance
Fourteenth Scandinavian Conference on Information Systems (SCIS2023), Porvoo, Finland 13
Greenwood, R., Raynard, M., Kodeih, F., Micelotta, E. R., & Lounsbury, M. (2011). Institutional Com-
plexity and Organizational Responses. Academy of Management Annals, 5(1), 317-371.
https://doi.org/10.5465/19416520.2011.590299
Hagendorff, T. (2020). The Ethics of AI Ethics: An Evaluation of Guidelines. Minds and Machines,
30(1), 99-120. https://doi.org/10.1007/s11023-020-09517-8
Hansen, S., & Baroody, A. J. (2020). Electronic health records and the logics of care: complementarity
and conflict in the U.S. healthcare system. Information Systems Research, 31(1), 57-75.
https://doi.org/10.1287/isre.2019.0875
Ibáñez, J. C., & Olmeda, M. V. (2022). Operationalising AI ethics: how are companies bridging the gap
between practice and principles? An exploratory study. AI & SOCIETY, 37(4), 1663-1687.
https://doi.org/10.1007/s00146-021-01267-0
Jaakkola, E. (2020). Designing conceptual articles: four approaches. AMS Review, 10(1-2), 18-26.
https://doi.org/10.1007/s13162-020-00161-0
Jobin, A., Ienca, M., & Vayena, E. (2019). The global landscape of AI ethics guidelines. Nature Machine
Intelligence, 1(9), 389-399. https://doi.org/10.1038/s42256-019-0088-2
Kaminski, M. E., & Malgieri, G. (2021). Algorithmic impact assessments under the GDPR: producing
multi-layered explanations. International Data Privacy Law, 11(2), 125-144.
https://doi.org/10.1093/idpl/ipaa020
Keymolen, E. (2023). Trustworthy tech companies: talking the talk or walking the walk. AI and Ethics.
https://doi.org/10.1007/s43681-022-00254-5
Kraatz, M. S., & Block, E. S. (2017). Institutional pluralism revisited. In R. Greenwood, C. Oliver, T.
B. Lawrence, & R. E. Meyer (Eds.), The Sage handbook of organizational institutionalism. SAGE
Inc.
Laato, S., Birkstedt, T., Mäantymäki, M., Minkkinen, M., & Mikkonen, T. (2022a). AI governance in
the system development life cycle. In Proceedings of the 1st International Conference on AI Engi-
neering: Software Engineering for AI. New York, NY, USA: ACM.
http://dx.doi.org/10.1145/3522664.3528598
Laato, S., Tiainen, M., Najmul Islam, A. K. M., & Mäntymäki, M. (2022b). How to explain AI systems
to end users: a systematic literature review and research agenda. Internet Research, 32(7), 1-31.
https://doi.org/10.1108/intr-08-2021-0600
Li, B., Qi, P., Liu, B., Di, S., Liu, J., Pei, J., Yi, J., & Zhou, B. (2023). Trustworthy AI: From Principles
to Practices. ACM Computing Surveys, 55(9), 1-46. https://doi.org/10.1145/3555803
Lim, W. M., Gunasekara, A., Pallant, J. L., Pallant, J. I., & Pechenkina, E. (2023). Generative AI and
the future of education: Ragnarök or reformation? A paradoxical perspective from management ed-
ucators. The International Journal of Management Education, 21(2), 100790.
https://doi.org/10.1016/j.ijme.2023.100790
Lukka, K., & Vinnari, E. (2014). Domain theory and method theory in management accounting research.
Accounting, Auditing & Accountability Journal, 27(8), 1308-1338. https://doi.org/10.1108/aaaj-03-
2013-1265
Mäntymäki, M., Minkkinen, M., Birkstedt, T., & Viljanen, M. (2022a). Defining organizational AI gov-
ernance. AI and Ethics, 2(4), 603-609. https://doi.org/10.1007/s43681-022-00143-x
Mäntymäki, M., Minkkinen, M., Birkstedt, T., & Viljanen, M. (2022b). Putting AI ethics into practice:
the hourglass model of organizational AI governance. http://arxiv.org/abs/2206.00335
Meske, C., Bunde, E., Schneider, J., & Gersch, M. (2022). Explainable Artificial Intelligence: Objec-
tives, Stakeholders, and Future Research Opportunities. Information Systems Management, 39(1),
53-63. https://doi.org/10.1080/10580530.2020.1849465
Metcalf, J., Moss, E., Watkins, E. A., Singh, R., & Elish, M. C. (2021). Algorithmic Impact Assessments
and Accountability. In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and
Transparency (pp. 735-746). New York, NY, USA: ACM.
http://dx.doi.org/10.1145/3442188.3445935
Minkkinen, M., Laine, J., & Mäntymäki, M. (2022a). Continuous Auditing of Artificial Intelligence: a
Conceptualization and Assessment of Tools and Frameworks. Digital Society, 1(3), 21.
https://doi.org/10.1007/s44206-022-00022-2
Institutional logics underpinning AI governance
Fourteenth Scandinavian Conference on Information Systems (SCIS2023), Porvoo, Finland 14
Minkkinen, M., Niukkanen, A., & Mäntymäki, M. (2022b). What about investors? ESG analyses as
tools for ethics-based AI auditing. AI & SOCIETY. https://doi.org/10.1007/s00146-022-01415-0
Minkkinen, M., Zimmer, M. P., & Mäntymäki, M. (2023). Co-Shaping an Ecosystem for Responsible
AI: Five Types of Expectation Work in Response to a Technological Frame. Information Systems
Frontiers, 25(1), 103-121. https://doi.org/10.1007/s10796-022-10269-2
Mittelstadt, B. (2019). Principles alone cannot guarantee ethical AI. Nature Machine Intelligence, 1(11),
501-507. https://doi.org/10.1038/s42256-019-0114-4
Mökander, J., Morley, J., Taddeo, M., & Floridi, L. (2021). Ethics-Based Auditing of Automated Deci-
sion-Making Systems: Nature, Scope, and Limitations. Sci Eng Ethics, 27(4), 44.
https://doi.org/10.1007/s11948-021-00319-4
Morley, J., Floridi, L., Kinsey, L., & Elhalal, A. (2020). From What to How: An Initial Review of
Publicly Available AI Ethics Tools, Methods and Research to Translate Principles into Practices.
Science and Engineering Ethics, 26(4), 2141-2168. https://doi.org/10.1007/s11948-019-00165-5
Morley, J., Kinsey, L., Elhalal, A., Garcia, F., Ziosi, M., & Floridi, L. (2023). Operationalising AI ethics:
barriers, enablers and next steps. AI & SOCIETY, 38(1), 411-423. https://doi.org/10.1007/s00146-
021-01308-8
O’Connor, M., Conboy, K., & Dennehy, D. (2023). Time is of the essence: a systematic literature review
of temporality in information systems development research. Information Technology & People,
36(3), 1200-1234. https://doi.org/10.1108/itp-11-2019-0597
Oborn, E., Pilosof, N. P., Hinings, B., & Zimlichman, E. (2021). Institutional logics and innovation in
times of crisis: Telemedicine as digital ‘PPE’. Information and Organization, 31(1), 100340.
https://doi.org/10.1016/j.infoandorg.2021.100340
Ocasio, W., Thornton, P. H., & Lounsbury, M. (2017). Advances to the institutional logics perspective.
In R. Greenwood, C. Oliver, T. B. Lawrence, & R. E. Meyer (Eds.), The Sage handbook of organi-
zational institutionalism (pp. 509-531). SAGE Inc.
OpenAI. (2023a). Introducing ChatGPT. https://openai.com/blog/chatgpt
OpenAI. (2023b). Product safety standards. https://openai.com/safety-standards
OpenAI. (2018). OpenAI Charter. https://openai.com/charter
Pache, A.-C., & Santos, F. (2010). When Worlds Collide: The Internal Dynamics of Organizational
Responses to Conflicting Institutional Demands. Academy of Management Review, 35(3), 455-476.
https://doi.org/10.5465/amr.35.3.zok455
Palladino, N. (2022). A ‘biased’ emerging governance regime for artificial intelligence? How AI ethics
get skewed moving from principles to practices. Telecommunications Policy, 102479.
https://doi.org/10.1016/j.telpol.2022.102479
Papagiannidis, E., Enholm, I. M., Dremel, C., Mikalef, P., & Krogstie, J. (2023). Toward AI Govern-
ance: Identifying Best Practices and Potential Barriers and Outcomes. Information Systems Fron-
tiers, 25(1), 123-141. https://doi.org/10.1007/s10796-022-10251-y
Radu, R. (2021). Steering the governance of artificial intelligence: national strategies in perspective.
Policy and Society, 40(2), 178-193. https://doi.org/10.1080/14494035.2021.1929728
Raji, I. D., Gebru, T., Mitchell, M., Buolamwini, J., Lee, J., & Denton, E. (2020). Saving Face: Inves-
tigating the Ethical Concerns of Facial Recognition Auditing. In Proceedings of the AAAI/ACM
Conference on AI, Ethics, and Society. New York, NY, USA: ACM.
http://dx.doi.org/10.1145/3375627.3375820
Resseguier, A., & Rodrigues, R. (2021). Ethics as attention to context: recommendations for the ethics
of artificial intelligence. https://open-research-europe.ec.europa.eu/articles/1-27/v1
Sætra, H. S. (2023). The AI ESG protocol: Evaluating and disclosing the environment, social, and gov-
ernance implications of artificial intelligence capabilities, assets, and activities. Sustainable Devel-
opment, 31(2), 1027-1037. https://doi.org/10.1002/sd.2438
Schiff, D., Rakova, B., Ayesh, A., Fanti, A., & Lennon, M. (2021). Explaining the Principles to Practices
Gap in AI. IEEE Technology and Society Magazine, 40(2), 81-94.
https://doi.org/10.1109/mts.2021.3056286
Institutional logics underpinning AI governance
Fourteenth Scandinavian Conference on Information Systems (SCIS2023), Porvoo, Finland 15
Schneider, J., Abraham, R., Meske, C., & Vom Brocke, J. (2022). Artificial Intelligence Governance
For Businesses. Information Systems Management, 1-21.
https://doi.org/10.1080/10580530.2022.2085825
Seger, E. (2022). In Defence of Principlism in AI Ethics and Governance. Philosophy & Technology,
35(2), 45. https://doi.org/10.1007/s13347-022-00538-y
Seppälä, A., Birkstedt, T., & Mäntymäki, M. (2021). From ethical AI principles to governed AI. In
ICIS. https://aisel.aisnet.org/icis2021/ai_business/ai_business/10
Shneiderman, B. (2020). Bridging the Gap Between Ethics and Practice. ACM Transactions on Interac-
tive Intelligent Systems, 10(4), 1-31. https://doi.org/10.1145/3419764
Smets, M., Morris, T., & Greenwood, R. (2012). From Practice to Field: A Multilevel Model of Practice-
Driven Institutional Change. Academy of Management Journal, 55(4), 877-904.
https://doi.org/10.5465/amj.2010.0013
Spataro, J. (2023). Introducing Microsoft 365 Copilot – your copilot for work. https://blogs.mi-
crosoft.com/blog/2023/03/16/introducing-microsoft-365-copilot-your-copilot-for-work/
Stahl, B. C., Antoniou, J., Ryan, M., Macnish, K., & Jiya, T. (2022). Organisational responses to the
ethical issues of artificial intelligence. AI & SOCIETY, 37(1), 23-37. https://doi.org/10.1007/s00146-
021-01148-6
Stix, C. (2022). The Ghost of AI Governance Past, Present, and Future. In J. Bullock & V. Hudson
(Eds.), The Oxford Handbook of AI Governance. Oxford University Press.
https://doi.org/10.1093/oxfordhb/9780197579329.013.56
Strümke, I., Slavkovik, M., & Madai, V. I. (2022). The social dilemma in artificial intelligence devel-
opment and why we have to solve it. AI and Ethics, 2(4), 655-665. https://doi.org/10.1007/s43681-
021-00120-w
Thiebes, S., Lins, S., & Sunyaev, A. (2021). Trustworthy artificial intelligence. Electronic Markets,
31(2), 447-464. https://doi.org/10.1007/s12525-020-00441-4
Thornton, P. H., & Ocasio, W. (2008). Institutional logics. In R. Greenwood, C. Oliver, R. Suddaby, &
K. Sahlin-Andersson (Eds.), The Sage handbook of organizational institutionalism (pp. 99-128).
Sage.
Thornton, P. H., Ocasio, W., & Lounsbury, M. (2012). The institutional logics perspective: a new ap-
proach to culture, structure, and process. Oxford University Press.
Tiwana, A., Konsynski, B., & Venkatraman, N. (2013). Special Issue: Information Technology and Or-
ganizational Governance: The IT Governance Cube. Journal of Management Information Systems,
30(3), 7-12. https://doi.org/10.2753/MIS0742-1222300301
Tournas, L. N., & Bowman, D. M. (2021). AI Insurance: Risk Management 2.0. IEEE Technology and
Society Magazine, 40(4), 52-56. https://doi.org/10.1109/mts.2021.3123750
Wei, M., & Zhou, Z. (2023). AI Ethics Issues in Real World: Evidence from AI Incident Database. In
56th Hawaii International Conference on System Sciences. Maui, Hawaii. https://hdl.han-
dle.net/10125/103236
Weill, P., & Ross, J. W. (2004). IT Governance: How Top Performers Manage IT Decision Rights for
Superior Results. Harvard Business Press.
http://books.google.fi/books?id=xI5KdR21QTAC&hl=&source=gbs_api
Whittlestone, J., Nyrup, R., Alexandrova, A., & Cave, S. (2019). The Role and Limits of Principles in
AI Ethics. In Proceedings of the 2019 AAAI/ACM Conference on AI, Ethics, and Society (pp. 195-
200). New York, NY, USA: ACM. http://dx.doi.org/10.1145/3306618.3314289
Yeung, K., Howes, A., & Pogrebna, G. (2020). AI Governance by Human Rights–Centered Design,
Deliberation, and Oversight: An End to Ethics Washing. In M. D. Dubber, F. Pasquale, & S. Das
(Eds.), The Oxford Handbook of Ethics of AI (pp. 75-106). Oxford University Press. https://ox-
fordhandbooks.com/view/10.1093/oxfordhb/9780190067397.001.0001/oxfordhb-9780190067397-
e-5
Zhu, L., Xu, X., Lu, Q., Governatori, G., & Whittle, J. (2021). AI and Ethics – Operationalising Re-
sponsible AI. In F. Chen & J. Zhou (Eds.), Humanity Driven AI. https://doi.org/10.1007/978-3-030-
72188-6_2
