ZAJKO - Artificial Intelligence Algorithms and Social Inequality Sociological Contributions To Contemporary Debates

Received: 19 June 2021 Revised: 19 January 2022 Accepted: 19 January 2022
DOI: 10.1111/soc4.12962
ARTICLE
Artificial intelligence, algorithms, and social

inequality: Sociological contributions to
contemporary debates
Mike Zajko
Department of History and Sociology, The

University of British Columbia Okanagan, Abstract
Kelowna, British Columbia, Canada
Artificial intelligence (AI) and algorithmic systems have been
Correspondence
criticized for perpetuating bias, unjust discrimination, and
Mike Zajko, Department of History and contributing to inequality. Artificial intelligence researchers
Sociology, The University of British Columbia
have remained largely oblivious to existing scholarship on
Okanagan, Kelowna, British Columbia, V1V
1V7, Canada. social inequality, but a growing number of sociologists are
Email: mike.zajko@ubc.ca
now addressing the social transformations brought about by
Funding information AI. Where bias is typically presented as an undesirable char-
University of British Columbia; Social Sciences acteristic that can be removed from AI systems, engaging
and Humanities Research Council of Canada,
Grant/Award Number: 430-2021-00810 with social inequality scholarship leads us to consider how
these technologies reproduce existing hierarchies and the
positive visions we can work towards. I argue that sociolo-
gists can help assert agency over new technologies through
three kinds of actions: (1) critique and the politics of refusal;
(2) fighting inequality through technology; and (3) govern-
ance of algorithms. As we become increasingly dependent
on AI and automated systems, the dangers of further en-
trenching or amplifying social inequalities have been well
documented, particularly with the growing adoption of
these systems by government agencies. However, public
policy also presents some opportunities to restructure social
dynamics in a positive direction, as long as we can articulate
what we are trying to achieve, and are aware of the risks and
This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits
use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or
adaptations are made.
© 2022 The Authors. Sociology Compass published by John Wiley & Sons Ltd.
Sociology Compass. 2022;16:e12962. wileyonlinelibrary.com/journal/soc4 1 of 16

https://doi.org/10.1111/soc4.12962
17519020, 2022, 3, Downloaded from https://compass.onlinelibrary.wiley.com/doi/10.1111/soc4.12962 by Readcube (Labtiva Inc.), Wiley Online Library on [31/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
2 of 16 ZAJKO
limitations of utilizing these new technologies to address so-

cial problems.
KEYWORDS
algorithms, artificial intelligence, bias, inequality, public policy,
technology
1 | INTRODUCTION
Over the past several years, the idea that society is being radically transformed through artificial intelligence (AI),
machine learning, and automation has been widely discussed in popular venues. This transformation has been ad-
dressed through a growing body of sociological scholarship (see Joyce et al., 2021; also; Airoldi, 2022; Burrell &
Fourcade, 2021; Davis et al., 2021; Elliott, 2022; Issar & Aneesh, 2022; Jaton, 2021), although some of the most
prominent sociological analyses of these technologies have come from academics who are not (formally) sociologists
(Crawford, 2021; Eubanks, 2017; Zuboff, 2018). Societal questions are now recognized as crucial ones for the de-
velopers of new algorithmic technologies, but ‘AI scientists continue to demonstrate a limited understanding of the
social’ (Joyce et al., 2021, p. 5). Much of the discourse addressing algorithmic technologies remains characterized by a
sense of inevitability and technological determinism (Vicsek, 2020), and one of the biggest concerns around the ‘rise
of AI’ has been a corresponding loss of human agency (Anderson & Rainie, 2018). This article will sketch out some
of the ways that sociologists can actively contribute to a better world, or oppose the dangerous visions that are so
common in this space.
I am particularly interested in how sociologists can approach a set of problems in the field of AI, machine learning,
and automated decision-making (ADM) that are predominantly understood as issues of ‘bias’ by technologists, but
which have roots in pre-existing social inequalities. Given the ongoing proliferation of algorithmic systems, sociolo-
gists can study these technologies and their unequal consequences in familiar domains such as work, education, and
law. However, to go beyond an empirical analysis of current developments, there are three kinds of contributions
sociologists can make through praxis, policy influence, or interdisciplinary collaboration: (1) critique and the politics
of refusal; (2) fighting inequality through technology; and (3) the governance of algorithmic governance. All three
should be considered important modes of positive action at a time when societies are being reconfigured through
algorithmic governance, and when the algorithmic reproduction of inequality is increasingly recognized as a major
issue. Critique and the politics of refusal represent a well-worn path for sociologists, through which we can also
articulate a positive vision for the world, even if our agency to pursue it remains limited. For more direct influence,
there are opportunities for sociologists to participate in the design and implementation of algorithmic systems to
improve social outcomes, but our participation can carry the risk of strengthening already-dominant interests. One
of the most important interdisciplinary contributions from sociology may be to unsettle and reconstitute established
problems in new ways.
2 | BACKGROUND
In much of the world, automated and algorithmic decision-making systems now govern various aspects of people's
lives, including those most directly implicated in social inequality, such as employment, government benefits, and
criminalization (Benjamin, 2019; Eubanks, 2017; O’Neil, 2016; Schuilenburg & Peeters, 2021). It is now increasingly
recognized that machine systems built to be objective and unbiased do indeed discriminate along familiar human lines,
reproducing or amplifying social differences and inequalities (Noble, 2018). The disposition of automated systems
ZAJKO 3 of 16
to uphold the existing social order has been described as a ‘conservative’ tendency (Birhane, 2020; Zajko, 2021), but
among AI researchers and developers, the general understanding is that people (or society) exhibit various ‘biases’
which are then reproduced in automated systems. This is particularly the case for today's dominant forms of AI or
‘machine learning’ algorithms, which must be ‘trained’ using datasets that reflect human judgments, priorities, and
conceptual categories (Crawford, 2021). When this data is biased, or when the underlying ‘ground truth’ of society
in unequal, this inequality can be encoded and reproduced in algorithms based on machine learning (Hacker, 2018).
Data about patterns in society serves as the input on which these systems are trained, with the resulting automated
decisions (the output) reflecting and perpetuating social inequalities. Machine learning and algorithmic governance
therefore frequently create ‘feedback loops’ that replicate and amplify existing patterns in society (Brayne, 2020;
Brayne & Christin, 2021; Mehrabi et al., 2019).
The solution to this problem among AI developers has been to find various ways to remove, reduce or ‘mini-
mize’ bias in datasets and algorithmic decisions (Silberg & Manyika, 2019). This is now increasingly recognized to
be a very complex and multi-dimensional challenge that cannot be achieved purely through technological solutions
(Kind, 2020). None of this has been surprising to scholars in fields such as science and technology studies (STS),
critical data studies, or critical algorithm studies, who have long examined the relationships between technology
and social structures (Felt et al., 2017; Gillespie & Seaver, 2016; Iliadis & Russo, 2016). Sociologists in general have
been less engaged in these issues thus far, which is unfortunate given that social theory can fill the conceptual chasm
currently identified as ‘bias’ in data science (Joyce et al., 2021; Zajko, 2021).
While some forms of bias in AI are roughly equivalent to how this term is understood in social research methods
and sampling (particularly given the statistical foundations for today's AI technologies), other biases in AI reflect what
would be understood by a sociologist as various kinds of inequality. These inequalities are sometimes categorized as
‘societal bias’ (Friedman & Nissenbaum, 1996) or ‘historical bias’ (Suresh & Guttag, 2020) by AI practitioners, which
has the effect of black-boxing any specifics about where the bias comes from or is structured. But sociologists can
name the intersecting structures (Davis et al., 2021; Hoffmann, 2019) of inequality, oppression, or exploitation that
produced the biases AI and data science practitioners are concerned about: colonialism, heteropatriarchy, ableism,
white supremacy, and various articulations of capitalism or political economy. While AI developers have been repeat-
edly criticized for being unwilling to look beyond their own field to ‘see the social’ (Irani & Chowdhury, 2019), this has
slowly been changing, and social scientists have been identified as a source of relevant expertise by AI researchers
(CIFAR, 2020; Kusner & Loftus, 2020), tech companies (Maris, 2022), academic bodies and governments (G7 Science
Academies, 2019). In short, there is an opportunity for sociologists to address pressing issues concerning the auto-
mation of inequality, by building on the field's long history of studying social stratification, power, oppression, and
intersectionality (Joyce et al., 2021).
3 | BIAS AS SOCIETY'S “UNEQUAL GROUND TRUTH”
In sociology, teaching students that society is ‘not fair’ is a regular part of the undergraduate curriculum. Media and
cultural studies have long documented inequalities in how different groups of people are represented, the exclusion
of groups, and how media privilege dominant categories (e.g., Erigha, 2015; Pascale, 2013). For developers of AI
systems however, the evident unfairness of algorithmic technologies (including blatantly racist or sexist algorithmic
outputs) has in recent years created major conceptual difficulties for which they were untrained and unprepared.
Solutions have been developed to minimize bias or increase fairness, but there is no consensus on what a fair or un-
biased algorithm might look like, or if this is even achievable (Green & Hu, 2018; Silberg & Manyika, 2019). Artificial
intelligence developers refer to the reality that exists outside of their models as the ‘ground truth’, and bias is often
defined as deviations from this truth, or inaccurate representations and predictions. But when the truth is that society
is deeply, structurally unjust and unequal, and that technologies are part of these structures, the question is whether
our algorithms should accurately reproduce inequality or work to change it.
4 of 16 ZAJKO
While the conventional approach in AI and data science treats bias as ‘error’ – inaccurate modeling or prediction – a
broader definition of bias that is also used (but less often articulated) equates it with whatever might be undesirable
in society (Mitchell et al., 2020; Zajko, 2021). In short, while an algorithmic model that accurately reproduces soci-
ety's ‘unequal ground truth’ might be considered unbiased in the first narrow sense of bias as accuracy, it may be
biased according to this second view of bias as an undesirable tendency. The problem for data science is that while
accuracy presents a clear goal to work towards through the elimination or reduction in bias, the second definition
of bias as an undesirable tendency does not specify what is desirable, beyond the contested notions of equality and
fairness (both of which have a number of competing and irreconcilable definitions, see Friedler et al., 2016; Silberg
& Manyika, 2019).
Essentially, what has happened in recent years is that AI scholarship has ‘discovered’ that social inequality is real.
This inequality consistently shows up in the data used to train algorithms through machine learning, as well as the
outputs of these systems. Practitioners have approached these problems through the language of bias, but bias is not
helpful in understanding how to deal with pre-existing inequalities that do not result from flaws in data or methods.
While there are many inequalities that are deemed to be generally problematic or undesirable, there is far less agree-
ment on what a desirable world looks like, and the aversion that many technologists feel towards taking a political
stance has also limited this discussion. Google, for instance, has explained offensive search results as accurate rep-
resentations of a ground truth (either the search behaviour of other users, or of online content – see Gibbs, 2016).
In situations where Google has modified search algorithms producing racist and sexist outputs, it has often done so
quietly, in response to media and public attention over specific cases (Noble, 2018), pressure from external interests,
or for internal reasons (West et al., 2019).
Whether we are looking at algorithmic search results, hiring decisions, or recidivism prediction, we will continue
to confront examples of harmful discrimination that cannot be reduced to issues of accuracy; any attempt at a solu-
tion will be inherently political, and avoiding the issue or treating it as being ‘out of scope’ for data science amounts to
a conservative orientation (Green, 2018; Zajko, 2021). As a consequence, a growing movement among AI researchers
has been taking a more radical path to altering relations of power (Kalluri, 2020), but interdisciplinary contributions
from scholars of power and inequality remain badly needed.
4 | SOCIAL INEQUALITY AND TECHNOLOGY: THE VIEW FROM SOCIOLOGY
Sociologists presume the existence of inequalities in any domain of society, and recognize that the benefits and
harms of technology are not evenly distributed. Scholarship on the ‘digital divide’ has examined unequal access and
use of technologies since the 1990s (van Dijk, 2006; Wellman & Haythornthwaite, 2002); when the Internet was
presented as a broadly empowering force for social good, sociologists warned that these technologies would also
reinforce numerous existing inequalities (DiMaggio et al., 2004). Selwyn (2004) and Helsper (2008, 2012) drew on
Bourdieusian theory and the concept of ‘social exclusion’ (see also van Dijk, 2005), to redefine the problem of the
digital divide as one of ‘digital exclusion’, with socio-technical inequalities ‘rooted in broader social categories linked
to other types of disadvantage and discrimination’ (Helsper, 2012, p. 406). Park and Humphry (2019) extended this
focus on exclusion to AI and ADM, analyzing how automated systems discriminate, deny, and punish welfare recipi-
ents in Australia (see also Eubanks, 2017; James & Whelan, 2021; Schou & Pors, 2019). Relatedly, sociologists of ed-
ucation (Davies et al., 2021; Williamson & Eynon, 2020) and medical technologies (Roberts & Rollins, 2020; Singh &
Steeves, 2020) have studied how algorithms exclude and discriminate on the basis of pre-existing social inequalities.
Sociologists of work, occupations, and organizations have examined how new algorithmic systems reconfig-
ure human labor, in ways that are often detrimental to workers (Bailey et al., 2020; Griesbach et al., 2019; Kellogg
et al., 2020; Newlands, 2021; Shestakofsky, 2017). Utopian and dystopian predictions of robots taking and trans-
forming human jobs have been the subject of discourse analysis (see James & Whelan, 2021; Ossewaarde & Gu-
lenc, 2020; Vicsek, 2020), but sociological scholarship has been skeptical or critical of these claims, and more at-
ZAJKO 5 of 16
tentive to questions of power relations (Boyd & Holton, 2018). Judy Wajcman contextualizes the latest wave of
‘hyperbole about AI’ (2017, p. 121) within ‘the perennial anxiety about automation’ (p. 125), reminding us to remain
focused on the concentration of power with a few large corporations and the biases of Silicon Valley's (typically male
and white) engineers. With the benefit of some historical perspective, our worries about the looming automation of
human labor is misplaced, when ‘the real issue is the unequal distribution of work, time and money that exist already’
and how new technologies are creating ‘not less work but more worse jobs’ (Wajcman, 2017, p. 124).
The inequalities that are reproduced and reshaped through algorithmic technologies can be studied within organ-
izations, but inequalities also play out on a global scale (Sampath, 2021), including international labor (Aneesh, 2009),
and the flow of capital through colonial and extractive processes (Couldry & Mejias, 2019). Sociology's own colo-
niality (Bhambra, 2014) has often excluded societies of the ‘Global South’ from analysis, or treated societies of the
North as universal (Go, 2020; Milan & Treré, 2019). While the most technologically-developed industrial nations in
North America, Europe, and East Asia are presented as the key competitors in the ‘race for AI’ (Walch, 2020), algo-
rithmic systems reach around the world and depend on global resources (Crawford, 2021). In the Global South, AI
systems have been promoted for international development (Sinanan & McNamara, 2021), but are often complicit in
processes of colonization and extraction (Couldry & Mejias, 2019; Kwet, 2019; Mohamed et al., 2020). Populations
in the Global South have a different relationship with major AI platforms than those who live and work where these
companies are headquartered (Aneesh, 2009; Gray & Suri, 2019; Grohmann & Araújo, 2021). Ideas from the Global
South (and Indigenous epistemologies in Northern societies) can also be the source of conceptual alternatives to
dominant theories and ethical rationalities of AI (Milan & Treré, 2019), including new ways to theorize, analyze, and
critique these socio-technical systems (Birhane, 2021; Lewis et al., 2020). Sociologists can play a valuable role in
bringing forward perspectives from social positions that have historically been marginalized in the development of AI.
5 | THE FUTURE OF INEQUALITY AND SOCIOLOGY'S RESPONSE
As the previous section outlined, numerous sociologists are working in domains where algorithmic and AI systems are
impossible to ignore, and carry significant consequences. Some sociologists have addressed AI in a general sense, ei-
ther to argue that it is indeed transforming our everyday lives (Elliott, 2019), or to push back against grandiose claims
of AI's capabilities (Collins, 2018). Burrell and Fourcade (2021) remind us that ‘we can both reject magical thinking
about machine intelligence and acknowledge the enormous economic, political, and cultural power of the tech indus-
try to transform the world we live in’ (p. 231). Questions of power (see Kalluri, 2020) should be a primary concern for
us as sociologists, given what is at stake in this ongoing transformation and the inequalities that algorithmic systems
can either shift or lock into place.
Projecting from current trends, it is easy to imagine a future world where wealth and power are even more con-
centrated in the hands of the ‘coding elite’ (Burrell & Fourcade, 2021) than today – a scenario that one RAND-affiliated
author has called ‘Bezos World’ (Lempert, 2019). This sort of speculative dystopia has already generated social en-
gineering proposals to redistribute wealth and inequality along both radical and conservative lines – from ‘luxury
communism’ (Bastani, 2019), to new capital taxes that fund basic income for citizens (Clifford, 2021), or a ‘cyber
republic’ in which new kinds of ownership and ‘data property rights’ are maintained through tokens and distributed
ledgers (Zarkadakis, 2020). These scenarios are typically presented as developing from an economic or political crisis
emerging as a consequence of widespread automation, whether this is imagined in the near future or as currently
underway. However, the social (re)engineering of inequality need not be contingent on some ‘new Industrial Revolu-
tion’ where robots replace human labor. Automated systems already distribute various goods and effects unequally,
or are complicit in the reproduction of long-established inequalities. The future, in this sense, promises more of the
same; control over sociotechnical systems will be used to further concentrate power along already-dominant lines.
Sociologists can theorize these developments by examining how social inequalities are structured, highlight-
ing political economy, capitalism, and colonial relations (Couldry & Mejias, 2019; Dyer-Witheford et al., 2019;
6 of 16 ZAJKO
Shestakofsky, 2020). While macro-level social theories provide analytic tools for global transformations, sociologists
can attend to the production of power and knowledge through genealogies (Denton et al., 2021) and ethnogra-
phies of AI research (Hoffman, 2021; Jaton, 2021). There will also be continuing value in producing ethnographies
(and institutional ethnographies, James & Whelan, 2021) of organizations implementing algorithmic systems (Bailey
et al., 2020; Brayne & Christin, 2021; Cruz, 2020; Shestakofsky & Kelkar, 2020), as well as studies into the experi-
ences of people who are further ‘downstream’, interacting with algorithmic systems (Christin, 2020; Noble, 2018).
Roberge and Castelle (2021) argue that we need an ‘end-to-end sociology’ of AI, investigating how these ‘upstream’
and ‘downstream’ processes are entangled, and tracing these sociotechnical systems ‘from genesis to impact and back
again’ (Roberge & Castelle, 2021, p. 3). Any analysis that takes a wider approach to the sociotechnical systems and
their relations also needs to consider the discourses and imaginary futures that are motivating these developments,
including the promotion of ideas about what kinds of futures are desirable or inevitable (James & Whelan, 2021;
Schuilenburg & Peeters, 2021; Vicsek, 2020).
Given that AI and algorithmic systems can be studied sociologically, how can sociology contribute to current
debates about these technologies? If the current trajectory leads to greater concentrations of wealth and power and
the reproduction of existing inequalities, how might we interrupt these developments? The loss of human agency
is a recurring worry in discussions of automation and AI (Anderson & Rainie, 2018), but we should recognize that
structured inequality has long deprived marginalized groups of certain forms of agency (Shah, 2018). The following
sections explain the different roles that sociologists can play in exercising our agency to change social structures and
shape a better world, by addressing algorithmically-mediated inequalities.
5.1 | Critique
Critique is a valuable mode of sociological argument that can analyze ‘how algorithms reproduce and reinforce exist-
ing structures’, situating new technologies within ‘broader political, racial, cultural, and economic formations’ (Chris-
tin, 2020, p. 900). In this regard, Ruha Benjamin's (2019) work has been exemplary (preceded notably by Virginia
Eubanks [2017] and Noble [2018]), as is that of many more interdisciplinary scholars in STS and related fields (Gil-
lespie & Seaver, 2016; Hoffmann, 2019). Critique helps unpack the politics of algorithmic technology, perhaps as
inspired by Winner (1980), or more contemporary examples from non-sociologists who engage with social theory
(Crawford, 2021; Zuboff, 2018). It is work that remains badly needed in this space, particularly in a form that is under-
standable by technologists for whom the language of power, politics, and ‘social impacts’ remains unfamiliar terrain
(CIFAR, 2020). Although writing across disciplines presents its own challenges within the field of sociology, critique
is an area of strength for us and can draw on many existing theories, skills, and methods.
While understanding the technical sophistication of AI can seem daunting if this is considered a prerequisite
for scholarly engagement, a critique of these systems can often begin by stating the sociologically obvious. Artificial
intelligence practitioners are continuously applying statistical methods to categorizing human populations, with very
little understanding of the social categories being operationalized. One recent study (since retracted due to lack of
ethics approval) cited 19th century criminologist Cesare Lombroso as a justification for identifying criminality through
facial features (Hashemi & Hall, 2020). Artificial intelligence practitioners continue to regularly produce work that can
be characterized as physiognomy or phrenology (Stark & Hutson, forthcoming; Stinson, 2021), classifying individuals
on the basis of discrete races, genders, and emotional states, through only the shallowest ontological engagement
with these phenomena (Barrett et al., 2019; Hanna et al., 2019; Scheuerman et al., 2021). This work typically demon-
strates little-to-no awareness of the pre-AI history of ‘social sorting’ (Lyon, 2003), and an inability to situate these sys-
tems in contemporary relations of governance or capital. Even though Silicon Valley-firms have employed significant
numbers of social scientists, humanities scholars, and even voices critical of the industry, the possibilities for critique
within these organizations are tenuous or circumscribed – as evidenced by Timnit Gebru's ouster from Google's AI
ZAJKO 7 of 16
Ethics team and the events that have followed (Grant et al., 2021; Simonite, 2021). Independent academic critique
remains badly needed from a variety of perspectives.
Critiques have sometimes been dismissed (particularly by practitioners) as having nothing to offer in terms of
recommendations or preferred policies. Many sociological critiques have indeed limited themselves to ‘unmasking
hidden structures of inequality’ (Cancian, 1995, p. 347), and stopped short of promoting social change or addressing
themselves to policymakers. But critique, rejection, and opposition can be the basis for various kinds of normative
arguments and practical ways forward. The ‘politics of refusal’ (Simpson, 2014) is at its core, a call to both oppose what
is wrong and to do things differently (see Cifor et al., 2019). As applied to algorithmic systems and AI (Crawford, 2021;
Eubanks, 2017), the politics of refusal is a more radical position than attempts to steer technologies in a more desir-
able direction, and it entails more than just refusing to contribute to the development of these systems. Rather, the
politics of refusal takes a normative stand against some aspects of algorithmic control, automation, data extraction,
and AI development. This opposition may be directed against specific applications of AI, such as facial recognition
and physiognomy (Stark & Hutson, forthcoming) or predictive policing (Roberts, 2019). Refusal of algorithmic tech-
nologies may also be one part of the larger project of dismantling unjust and violent institutions, such as the ‘carceral
state’ (Roberts, 2019), ‘digital poorhouse’ (Eubanks, 2017), and colonial structures (Walter et al., 2020). Arguments for
refusal are sometimes part of a more general critique of technological determinism and inevitability (Benjamin, 2019;
Crawford, 2021). Technologists and a broader public need to be regularly reminded that just because we can build or
automate something, does not mean that we should, and that rejection of the latest ‘innovation’ in social control is a
perfectly reasonable option.
Like the alternative approaches outlined below, the politics of refusal can be considered an exercise of human
agency over sociotechnical systems, and like these alternative approaches it should also be understood in positive
terms. As theorized by Audra Simpson (2014) and those who have built on her work, refusal is ‘generative’ (see
McGranahan, 2016) and a means to affirm who we are, our relations, and our values. Arguing that there are certain
decisions which must not be automated, contains within it a positive argument for what certain decisions should
entail, in terms of human involvement. Because of this, arguing that we should not build or implement a system is
valuable and often appropriate as a way to achieve desirable futures.
5.2 | Fighting inequality through technology
Structured inequality is reproduced on an ongoing basis and resistant to change, but the behaviour of digital systems
is written in code, and code can be re-written. In other words, when inequalities are encoded in algorithms, software,
or spreadsheets, changing the code will redistribute the inequalities. Research that addresses fairness and bias in data
science is already trying to change the world by reducing bias, and algorithmic governance is delivering new ways to
configure, engineer, and structure society. It is necessary to reject or resist some of these developments, but we can
also try to steer them away from their darkest possibilities, towards a world that better reflects our values.
Given that work on fairness, ethics, and bias in AI is now part of the ‘normative construction of the world’ (Green
& Hu, 2018, p. 5), sociologists can contribute to this world-building (Joyce et al., 2021) and help articulate possible
directions for change that go beyond nebulous notions of an unbiased society or undesirable tendencies. Where
vague gestures towards a ‘social good’ have predominated in AI ethics, there is an opportunity to specify desirable
futures based on substantive equality and anti-oppression (Green, 2019; also; Davis et al., 2021). Unfortunately,
while sociologists are on comfortable ground when critiquing representational inequalities and algorithmic harms, we
are less comfortable in making normative proposals for what to do with algorithmic systems, beyond rejecting their
use in problematic cases. A notable exception is Benjamin's (2019) Race After Technology, which does conclude with
sections on ‘abolitionist tools’ and ‘reimagining technology’, but the book remains largely a work of explanation and
critique and is very skeptical of ‘design thinking’. Benjamin reminds us that when we are dealing with inequalities in-
volving new technologies, we do not necessarily need a technological fix. Established forms of politics are often more
8 of 16 ZAJKO
appropriate, as expressed in the argument that ‘maybe what we must demand is not liberatory designs but just plain
old liberation’ (p. 179). Keeping this in mind, what would it mean to enact systemic reforms that address inequality, in
ways that go beyond rejection, refusal or opposition to sociotechnical systems? In other words, what positive actions
are possible (involving both technology and politics) to create the sort of world we want?
While many students are attracted to sociology because of a desire for social change, those of us em-
ployed in academia often have other priorities, as reflected by the work we do and its audiences (Cancian, 1995;
J. H. Turner, 2001; Weinstein, 2000). However, sociology does have a long history of arguing for social ‘improvement’
(Ward, 1906) and carrying out applied scholarship (Perlstadt, 2007). Along these lines, sociologists may engage in
applied or participatory work with specific groups that are exercising agency through technology and data in order
to meet their needs, or communities working towards some articulation of justice (Costanza-Chock, 2020). There are
many groups whose interests have been marginalized in the design of new technologies, and their struggles are not
as straight-forward as a fight for equality – encompassing goals that are specific to those communities. For example,
in the context of settler colonialism (see Simpson, 2014), treating everyone on equal terms amounts to a form of as-
similation for Indigenous peoples. The movement towards decolonization and Indigenous data sovereignty (Kukutai &
Taylor, 2016; Lewis et al., 2020; Walter et al., 2020) is therefore not based on liberal concepts of individual equality,
but the distinctive position of Indigenous peoples in regard to land rights, sovereignty, and cultural integrity.
Although there is a great deal of work that can be done to address the needs and desires of specific populations
in relation to technologies, the fact that algorithmic systems are now implicated in inequalities ‘at scale’ also presents
possibilities for more ambitious interventions. A national-level tax regime is one example of how a low-tech algo-
rithm can broadly shift lines of inequality, but there are many more specific regimes for the distribution of wealth
and resources that are in the process of being automated or ‘transformed’ through digital government (Clarke, 2020;
Henman, 2010; Levy et al., 2021). Additionally, there are the algorithmic systems operated by massive platform
companies (sometimes employing sociologists) that act as de facto governments, shaping outcomes for millions or
billions of people (Gillespie, 2018; Tusikov, 2017). Sociologists can certainly seek to contribute to these algorithmic
endeavors, ideally in an interdisciplinary manner, and mindful of the history of the relationship between sociology
and technocracy, or social engineering. While partnering with established centres of power brings greater capacities
to bear on the problems of algorithmic governance and inequality, the risk is that such work becomes subordinated
to dominant interests and may further entrench the status quo.
5.2.1 | Interdisciplinary social engineering
Social engineering is a term that has typically been applied to top-down and expert-led projects to order society,
particularly through state agencies. It can be exemplified by the policies of totalitarian governments of the twentieth
century, including Nazi and communist regimes, but also programs in more socially democratic as well as liberal states,
such as the eugenic policies that were once widespread in Europe and North America (Bashford & Levine, 2010;
Lucassen, 2010). The relationship between sociology and social planning was a significant theme in the early years
of the discipline (Ward, 1906), and while enthusiasm for social engineering faded in American social science over the
1930s (Jordan, 1994), it was addressed in the later work of Karl Mannheim (1940), as well as the relatively obscure
discipline of ‘sociotechnics’, associated with Polish sociologist Adam Podgórecki (Podgórecki et al., 1996).
Today, few in North America would characterize themselves as ‘social engineers’, but in Sweden social engi-
neering describes the establishment of the post-war welfare state. Although in Sweden the approach also largely
fell out of favor later in the twentieth century, it remains in use to refer to ‘the idea that we can create the great-
est degree of happiness, the ‘good society,’ through rational social planning’ (Etzemüller, 2014, p. 7). The develop-
ment of social engineering has much in common with the birth of technocracy (Jordan, 1994), which also originated
as an engineering approach to social problems, and uses specialized expertise or rational planning as a basis for
decision-making (Esmark, 2020). While technocracy is likewise much-critiqued as a form of governance (particularly
ZAJKO 9 of 16
for its anti-democratic orientation, see Fischer, 1990) and there are few who would self-identify as technocrats,
a technocratic rationality characterizes how many contemporary forms of governance operate across a variety of
political systems (Esmark, 2017, 2020). Meanwhile, it has been claimed that through the potential of algorithmic
technologies to reconfigure societies, ‘software engineers will increasingly be the social engineers of the digital
lifeworld’ (Susskind, 2018, p. 294).
The future relationship between sociology and social engineering has been discussed by Jonathan Turner (2001,
p. 101), who argued that sociology should develop an ‘engineering wing’ based on theoretical principles applied to
real-world problems, and an ability to translate knowledge to ‘fellow engineers’ and ‘clients’. Duncan J. Watts (2017)
made a related argument as a sociologist employed at Microsoft, and this is also the kind of approach favored by a
number of AI practitioners, imagining social scientists as problem solvers who can ‘propose, implement, and evaluate’
algorithmic processes to advance the ‘social good’ (Lepri et al., 2018). Data scientists and AI practitioners are more
likely to look to social science for solutions to an existing problem, than to consider how problems or perspectives
in social science translate to data science. For example, data scientists may have questions about human behaviour
or cognition that they imagine social scientists can design experiments to answer (Irving & Askell, 2019). Collabora-
tions between data science and social science around a specific topic or shared problem often end up being led by
technically-skilled participants, with programmers deciding what counts as important, while less-technical participants
take a back seat. This produces familiar ‘disciplinary divisions of labor: social scientists observe, data scientists make; so-
cial scientists do ethics, data scientists do science’ (Moats & Seaver, 2019, p. 8). In these processes, disciplinary bound-
aries are maintained and social science plays a supporting or ‘additive’ (Lury, 2018, p. 1) role in contributing knowledge.
In contrast to multidisciplinary work as described above, interdisciplinarity can be considered more of a recip-
rocal, transformative, or integrative relationship between disciplines. According to Lury, interdisciplinary methods
are ‘dynamic conduits for relations of interference in which differences and asymmetries between disciplines are
explored and exploited in relation to specific problems, in specific places, with specific materials’ (2018, p. 21). Here,
the emphasis is on crossing disciplines to ‘constitute’ problems in new ways, rather than arriving at novel solutions,
and interdisciplinarity is less an instrumental practice and more of an enterprise in research autonomy (Lury, 2018).
Even in situations where two disciplines use different language to discuss what might appear to be the same prob-
lem, switching from one disciplinary discourse to another can significantly shift how problems are formulated. This
is certainly the case when it comes to questions of bias and fairness in AI, once we attempt to translate these into
discourses of social inequality. ‘Reducing bias’ is not at all synonymous with ‘reducing oppression’. Researchers can
attempt to correct for ‘societal bias’ (Friedman & Nissenbaum, 1996; Silberg & Manyika, 2019), but understanding the
source of this structural inequality is typically treated as irrelevant, or outside the scope of analysis. Social structures,
or ‘sources of discrimination that cannot be traced to discrete bad mechanisms are bracketed, dismissed as someone
else's problem or, worse, couched as untouchable facts of history’ (Hoffmann, 2019, p. 905). Sociologists can play a
valuable role in the formulation of problems as well as in identifying where interventions may be effective, but inter-
disciplinary openness will be required from collaborators to make this possible.
For AI researchers and data scientists, the key problem in recent years has been eliminating unfair discrimination
and producing ‘fair’ results through automated decisions. If sociologists are brought in to help once a sociotechnical
system has already been proposed and the problem of fair decision-making raised, the scope of our agency is likely
to be quite limited. Sociologists may have more experience with philosophy and politics than those working in tech-
nical fields, but this does not confer an ability to configure a ‘fair’ distribution from an established decision-making
system. Sociology's strengths include our theoretical and empirical analyses of inequality – an understanding of
systems of stratification and forms of discrimination. Systemic problems such as racism and poverty are structurally
reproduced and require systemic approaches; sociologists can help to articulate what these approaches might look
like (e.g., Davis et al.’s [2021] proposal for ‘algorithmic reparation’), or where to focus our energies to achieve social
change. Therefore, the greatest contributions that sociologists can make are at earlier stages in the development of
policies and sociotechnical systems, where goals and techniques are less clearly defined. Once plans are underway,
10 of 16 ZAJKO
sociologists are likely to be enrolled in ways that are complicit with organizational goals such as efficiency, legitimacy,
or profit (Maris, 2022).
5.3 | Governance of algorithmic governance
Issues related to social inequality are of collective or public interest, and therefore are often pursued through insti-
tutions responsible for the public good. This leads to the third way that sociologists can be involved in the normative
construction of this world, by participating in the governance of technology, and addressing the very real dangers of
algorithmic governance.
Recent years have seen a considerable amount of scholarship documenting, comparing and critiquing different
policies or regimes for regulating algorithms and AI, including government strategies, corporate statements of princi-
ples, standards, and regulations (Bradley et al., 2020; Jobin et al., 2019; Stark et al., 2021; J. Turner, 2019). A positive
approach builds on this work to develop and improve these governance regimes. Seyfert (2021) argues that regu-
latory processes co-produce the algorithms they regulate, but AI-specific regulations are a new development, often
lacking ‘teeth’ or meaningful enforcement. To the extent that formal political channels are open to us, sociologists
can contribute to the development of the nascent regulatory regimes being established to govern AI and algorithms
in their respective jurisdictions or in an international forum.
The degree to which sociologists can contribute to public policy on these issues depends on the political oppor-
tunities available, such as whether formal processes are open to academic input, or the extent to which consultations
are actually used to inform policy (rather than playing a legitimating or performative role, see Kerr et al., 2020). Crit-
ical experts who are excluded from the policy process can establish their own organizations, as a number of leading
scholars did in 2020 with the ‘Real Facebook Oversight Board’ (Solon, 2020) – but these efforts remain at the level
of policy critique rather than policy making, and can be ignored by industry. The current moment shows a strong
appetite for government regulation of ‘tech giants’ in much of the world (Lee, 2021), and sociologists can contribute
by emphasizing questions of power, political economy and structural inequalities in these debates. While regula-
tors grapple with addressing the power that the corporate leaders in AI have achieved in recent years, the greatest
societal ‘disruptions’ will likely come through the use of these technologies by government agencies (CSPS, 2021).
Although the algorithms deployed by Google, Amazon, or Facebook affect billions of people, algorithms used in the
public sector can have more profound consequences – and are also more open to contestation and study by inde-
pendent researchers.
Sociologists need to study these emerging uses of AI in public institutions, while also fighting to keep them
publicly accountable. I have been documenting the use of ADM systems in Canada, where government agencies
have been employing these technologies for hiring decisions, benefits claims, immigration applications, legal analysis,
sentiment analysis, suicide prediction, fraud detection, facial recognition, and public-serving chatbots (among many
other uses – see ESDC, 2019; PWGSC, 2018; Reevely, 2021). Governing such algorithmic systems may include the
creation of new regulatory processes and agencies, but AI and algorithms are already regulated through privacy and
data protection law, competition law, human rights and anti-discrimination law, and more specialized domains where
these technologies are applied, such as medicine and public administration. Currently, algorithmic and AI regulations
have more to do with questions of transparency, accountability, privacy, competition, and fairness or justice, than
social inequality as a policy problem, but this could be changed by orienting current issues in AI regulation toward the
pursuit of substantive equality (Amani, 2021). Critical and engaged scholarship is needed both at the level of broad
regulatory regimes, as well as the specific implementation of these sociotechnical systems in particular domains,
which may have their own specific channels for democratic accountability and public participation.
Algorithmic governance therefore presents sociologists with a research problem, and a way to make a posi-
tive contribution to policy. Academic involvement can include carefully documenting algorithmic processes, how
they affect human lives and outcomes, critiquing and opposing the harms they cause, and reforming, improving, or
ZAJKO 11 of 16
constraining these systems through regulatory and public policy interventions. Broadly speaking, sociologists can
help define the role of automated systems in a democratic society. In comparison to twentieth-century social en-
gineering projects, today's technocratic ‘innovations’ are more frequently presented as compatible with (and even
bolstering) democratic processes (Esmark, 2020), but the historical critiques of technocracy in a democratic context
(Jordan, 1994) remain relevant, including the reliance on experts to determine the public interest and the dangers of
an ideological pursuit of efficiency. The use of automated systems to implement public policy in a democracy raises
major issues that require closer attention from all citizens, sociologists included.
6 | CONCLUSION
Sociologists must now grapple with the role of algorithmic systems and AI in the distribution of various goods and
outcomes; these systems are widespread today, and will continue to become even more central to the reproduction
of inequality in the future. Given the enormity of the ethical and political challenges involved, and the limitations that
other fields have demonstrated in addressing them, sociologists have an opportunity to make a valuable contribution
to human welfare. Specifically, issues that are currently presented as problems with bias can often be understood
through social theory as problems of inequality, made manifest in structures that reproduce benefits for some and
harms or cumulative disadvantages for others.
Sociologically-informed critiques of technologies have already had some influence on current debates but remain
in need of further elaboration. When critique extends into the politics of refusal, alliances can be formed with others
who are motivated to reject specific developments and promote more positive ideas for social change, including
alternative sociotechnical systems that are compatible with our visions and values. This might mean working with
groups and communities that are building technologies to meet their needs from the bottom-up, but it can also mean
critically engaging with systems explicitly intended to serve the public good – namely through the public sector.
Relatedly, broader democratic and interdisciplinary participation is needed in the development of new public policies
and governance mechanisms that will help co-produce the algorithms of the future.
This will require a practical orientation from sociologists, working beyond our discipline, and articulating nor-
mative positions and policy proposals that others will hear. Thankfully, much of the heavy lifting has already been
done by critical voices behind the ‘techlash’ of recent years. Algorithmic harms, racist and sexist robots, ubiquitous
surveillance and behavioral manipulation have now all become well-established as public issues. Artificial intelligence
researchers sometimes assume that social scientists can help them address these issues, and sociologists can indeed
choose to play such a collaborative role, as long as we are mindful of the limits of our agency and the danger of our
contributions becoming subordinated to other interests. Active engagement with algorithmic systems does not ne-
cessitate sociologists who can write code, but it does force us to consider the discipline's relationship to governance
and politics, including where we have the greatest possibilities of effecting change. Just as sociological critique can
be effective by locating technologies and individualized consequences in systemic terms or a larger context, applied
sociology may help orient these sociotechnical systems to address larger, interrelated, and systemic problems, includ-
ing various kinds of inequality and their consequences.
ACKNOWLE DG ME NT
None.
O RC ID
Mike Zajko https://orcid.org/0000-0001-7804-4618
12 of 16 ZAJKO
R EF ERE NCE S
Airoldi, M. (2022). Machine habitus: Toward a sociology of algorithms. Polity Press.
Amani, B. (2021). AI and equality by design. In F. Martin-Bariteau & T. Scassa (Eds.), Artificial intelligence and the law in Canada
(pp. 267–300).
Anderson, J., & Rainie, L. (2018). Artificial intelligence and the future of humans. Pew Research Center. https://www.pewre-
search.org/internet/2018/12/10/artificial-intelligence-and-the-future-of-humans/
Aneesh, A. (2009). Global labor: Algocratic modes of organization. Sociological Theory, 27(4), 347–370. https://doi.
org/10.1111/j.1467-9558.2009.01352.x
Bailey, S., Pierides, D., Brisley, A., Weisshaar, C., & Blakeman, T. (2020). Dismembering organisation: The coordination of algo-
rithmic work in healthcare. Current Sociology, 68(4), 546–571. https://doi.org/10.1177/0011392120907638
Barrett, L. F., Adolphs, R., Marsella, S., Martinez, A. M., & Pollak, S. D. (2019). Emotional expressions reconsidered: Challenges
to inferring emotion from human facial movements. Psychological Science in the Public Interest, 20(1), 1–68. https://doi.
org/10.1177/1529100619832930
Bashford, A. & Levine, P. (Eds.). (2010). The Oxford handbook of the history of eugenics. Oxford University Press.
Bastani, A. (2019). Fully automated luxury communism: A manifesto. Verso Books.
Benjamin, R. (2019). Race after technology: Abolitionist tools for the new Jim code. Polity Press.
Bhambra, G. K. (2014). Knowledge production in global context: Power and coloniality [Special issue]. Current Sociology, 62(4).
Birhane, A. (2020). Fair warning. Real life. https://reallifemag.com/fair-warning/
Birhane, A. (2021). Algorithmic injustice: A relational ethics approach. Patterns, 2(2), 1–9. https://doi.org/10.1016/j.
patter.2021.100205
Boyd, R., & Holton, R. J. (2018). Technology, innovation, employment and power: Does robotics and artificial intelligence
really mean social transformation? Journal of Sociology, 54(3), 331–345. https://doi.org/10.1177/1440783317726591
Bradley, C., Wingfield, R., & Metzger, M. (2020). National artificial intelligence strategies and human rights: A review.
https://www.gp-digital.org/wp-content/uploads/2020/04/National-Artifical-Intelligence-Strategies-and-Hu-
man-Rights%E2%80%94A-Review_April2020.pdf
Brayne, S. (2020). Predict and surveil: Data, discretion, and the future of policing. Oxford University Press.
Brayne, S., & Christin, A. (2021). Technologies of crime prediction: The reception of algorithms in policing and criminal courts.
Social Problems, 68, 608–624.
Burrell, J., & Fourcade, M. (2021). The society of algorithms. Annual Review of Sociology, 47(1), 213–237. https://doi.
org/10.1146/annurev-soc-090820-020800
Cancian, F. M. (1995). Truth and goodness: Does the sociology of inequality promote social betterment? Sociological Perspec-
tives, 38(3), 339–356. https://doi.org/10.2307/1389431
Christin, A. (2020). The ethnographer and the algorithm: Beyond the black box. Theory and Society, 49(5), 897–918.
CIFAR. (2020). DLRL summer school 2020—Ethics in AI - panel discussion. https://www.youtube.com/watch?v=ThYslOokwR4
Cifor, M., Garcia, P., Cowan, T. L., Rault, J., Sutherland, T., Chan, A. S., Rode, J., Hoffmann, A. L., Salehi, N., & Nakamura, L.
(2019). Feminist data manifest-no. https://www.manifestno.com/home
Clarke, A. (2020). Digital government units: What are they, and what do they mean for digital era public management
renewal? International Public Management Journal, 23(3), 358–379. https://doi.org/10.1080/10967494.2019.1686447
Clifford, C. (2021). OpenAI’s Sam Altman: Artificial Intelligence will generate enough wealth to pay each adult $13,500 a year.
CNBC. https://www.cnbc.com/2021/03/17/openais-altman-ai-will-make-wealth-to-pay-all-adults-13500-a-year.html
Collins, H. (2018). Artifictional intelligence: Against humanity’s surrender to computers. Polity Press.
Costanza-Chock, S. (2020). Design justice: Community-led practices to build the worlds we need. MIT Press.
Couldry, N., & Mejias, U. A. (2019). The costs of connection: How data is colonizing human life and appropriating it for capitalism.
Stanford University Press.
Crawford, K. (2021). Atlas of AI. Yale University Press.
Cruz, T. M. (2020). Perils of data-driven equity: Safety-net care and big data’s elusive grasp on health inequality. Big Data &
Society, 7(1), 1–14. https://doi.org/10.1177/2053951720928097
CSPS (2021). CSPS virtual café series: Artificial intelligence. https://www.youtube.com/watch?v=6drggKeFfC4
Davies, H. C., Eynon, R., & Salveson, C. (2021). The mobilisation of AI in education: A bourdieusean field analysis. Sociology,
55(3), 539–560. https://doi.org/10.1177/0038038520967888
Davis, J. L., Williams, A., & Yang, M. W. (2021). Algorithmic reparation. Big Data & Society, 8(2), 1–14. https://doi.
org/10.1177/20539517211044808
Denton, E., Hanna, A., Amironesei, R., Smart, A., & Nicole, H. (2021). On the genealogy of machine learning datasets: A critical
history of ImageNet. Big Data & Society, 8(2), 1–14. https://doi.org/10.1177/20539517211035955
DiMaggio, P., Hargittai, E., Celeste, C., & Shafer, S. (2004). Digital inequality: From unequal access to differentiated use. In K. M.
Neckerman (Ed.), Social inequality (pp. 355–400). Russell Sage Foundation. https://doi.org/10.7758/9781610444200.14
Dyer-Witheford, N., Kjøsen, A. M., & Steinhoff, J. (2019). Inhuman power: Artificial intelligence and the future of capitalism.
Pluto Press.
ZAJKO 13 of 16
Elliott, A. (2019). The culture of AI: Everyday life and the digital revolution. Routledge.
Elliott, A. (Ed.). (2022). The Routledge social science handbook of AI. Routledge.
Erigha, M. (2015). Race, gender, Hollywood: Representation in cultural production and digital media’s potential for change.
Sociology Compass, 9(1), 78–89. https://doi.org/10.1111/soc4.12237
ESDC. (2019). Final or most recent version of Employment and Social Development Canada artificial intelligence strategy (access
to information request A-2019-01027).
Esmark, A. (2017). Maybe it is time to rediscover technocracy? An old framework for a new analysis of administrative reforms
in the governance era. Journal of Public Administration Research and Theory, 27(3), 501–516. https://doi.org/10.1093/
jopart/muw059
Esmark, A. (2020). The new technocracy. Bristol University Press.
Etzemüller, T. (2014). Alva and Gunnar Myrdal: Social engineering in the modern world. Lexington Books.
Eubanks, V. (2017). Automating inequality: How high-tech tools profile, police, and punish the poor. Martin’s Press.
Felt, U., Fouché, R, Miller, C. A., & Smith-Doerr, L. (Eds.). (2017). The handbook of science and technology studies (4th ed.). MIT
Press.
Fischer, F. (1990). Technocracy and the politics of expertise. Sage.
Friedler, S. A., Scheidegger, C., & Venkatasubramanian, S. (2016). On the (im)possibility of fairness. ArXiv:1609.07236 [Cs, Stat].
http://arxiv.org/abs/1609.07236
Friedman, B., & Nissenbaum, H. (1996). Bias in computer systems. ACM Transactions on Information Systems, 14(3), 330–347.
https://doi.org/10.1145/230538.230561
Gibbs, S. (2016). Google alters search autocomplete to remove “are Jews evil” suggestion. The Guardian. https://www.theguardi-
an.com/technology/2016/dec/05/google-alters-search-autocomplete-remove-are-jews-evil-suggestion
Gillespie, T. (2018). Custodians of the internet: Platforms, content moderation, and the hidden decisions that shape social media.
Yale University Press.
Gillespie, T., & Seaver, N. (2016). Critical algorithm studies: A reading list. Social media collective. http://socialmediacollective.
org/reading-lists/critical-algorithm-studies/
Go, J. (2020). Race, empire, and epistemic exclusion: Or the structures of sociological thought. Sociological Theory, 38(2),
79–100. https://doi.org/10.1177/0735275120926213
Grant, N., Bass, D., & Eidelson, J. (2021). Google Ethical AI group’s turmoil began long before public unraveling. Bloomberg.
https://www.bloomberg.com/news/articles/2021-04-21/google-ethical-ai-group-s-turmoil-began-long-before-
public-unraveling
Gray, M. L., & Suri, S. (2019). Ghost work: How to stop Silicon Valley from building a new global underclass. Houghton Mifflin
Harcourt.
Green, B. (2018). Data science as political action: Grounding data science in a politics of justice. ArXiv Preprint ArXiv:1811.03435.
https://arxiv.org/abs/1811.03435
Green, B. (2019). “Good” isn’t good enough. Proceedings of the AI for social good workshop at NeurIPS. https://www.ben-
zevgreen.com/wp-content/uploads/2019/11/19-ai4sg.pdf
Green, B., & Hu, L. (2018). The myth in the methodology: Towards a recontextualization of fairness in machine learning. 35th inter-
national conference on machine learning. https://scholar.harvard.edu/files/bgreen/files/18-icmldebates.pdf
Griesbach, K., Reich, A., Elliott-Negri, L., & Milkman, R. (2019). Algorithmic control in platform food delivery work. Socius, 5,
2378023119870041. https://doi.org/10.1177/2378023119870041
Grohmann, R., & Araújo, W. F. (2021). Beyond Mechanical Turk: The work of Brazilians on global AI platforms. In P. Verdegem
(Ed.), AI for everyone? Critical perspectives (pp. 247–266). University of Westminster Press.
Hacker, P. (2018). Teaching fairness to artificial intelligence: Existing and novel strategies against algorithmic discrimination
under EU law. Common Market Law Review, 55(4), 1143–1185.
Hanna, A., Denton, E., Smart, A., & Smith-Loud, J. (2019). Towards a critical race methodology in algorithmic fairness. Conference
on Fairness, Accountability, and Transparency (FAT* ’20). https://doi.org/10.1145/3351095.3372826
Hashemi, M., & Hall, M. (2020). Criminal tendency detection from facial images and the gender bias effect. Journal of Big Data,
7(2), 1–16. https://doi.org/10.1186/s40537-019-0282-4
Helsper, E. J. (2008). Digital inclusion: An analysis of social disadvantage and the information society. Department for Commu-
nities and Local Government. http://www.communities.gov.uk/documents/communities/pdf/digitalinclusionanalysis
Helsper, E. J. (2012). A corresponding fields model for the links between social and digital exclusion. Communication Theory,
22(4), 403–426. https://doi.org/10.1111/j.1468-2885.2012.01416.x
Henman, P. (2010). Governing electronically e-government and the reconfiguration of public administration, policy and power.
Palgrave Macmillan.
Hoffman, S. G. (2021). A story of nimble knowledge production in an era of academic capitalism. Theory and Society, 50(4),
541–575. https://doi.org/10.1007/s11186-020-09422-0
Hoffmann, A. L. (2019). Where fairness fails: Data, algorithms, and the limits of antidiscrimination discourse. Information,
Communication & Society, 22(7), 900–915. https://doi.org/10.1080/1369118X.2019.1573912
14 of 16 ZAJKO
Iliadis, A., & Russo, F. (2016). Critical data studies: An introduction. Big Data & Society, 3(2), 1–7. https://doi.org/10.1177/
2053951716674238
Irani, L., & Chowdhury, R. (2019). To really “disrupt,” tech needs to listen to actual researchers. Wired. https://www.wired.com/
story/tech-needs-to-listen-to-actual-researchers/
Irving, G., & Askell, A. (2019). AI safety needs social scientists. Distill, 4. https://doi.org/10.23915/distill.00014
Issar, S., & Aneesh, A. (2022). What is algorithmic governance? Sociology Compass, 16(1), e12955. https://doi.org/10.1111/
soc4.12955
James, A., & Whelan, A. (2021). ‘Ethical’ artificial intelligence in the welfare state: Discourse and discrepancy in Australian
social services. Critical Social Policy, 42, 1–42. https://doi.org/10.1177/0261018320985463
Jaton, F. (2021). The constitution of algorithms: Ground-truthing, programming, formulating. MIT Press.
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
Jordan, J. M. (1994). Machine-age ideology: Social engineering and American liberalism, 1911-1939. University of North Carolina
Press.
Joyce, K., Smith-Doerr, L., Alegria, S., Bell, S., Cruz, T., Hoffman, S. G., Noble, S. U., & Shestakofsky, B. (2021). Toward a
sociology of artificial intelligence: A call for research on inequalities and structural change. Socius, 7, 1–11. https://doi.
org/10.1177/2378023121999581
Kalluri, P. (2020). Don’t ask if artificial intelligence is good or fair, ask how it shifts power. Nature, 583(7815), 169. https://doi.
org/10.1038/d41586-020-02003-2
Kellogg, K. C., Valentine, M. A., & Christin, A. (2020). Algorithms at work: The new contested terrain of control. The Academy
of Management Annals, 14(1), 366–410.
Kerr, A., Barry, M., & Kelleher, J. D. (2020). Expectations of artificial intelligence and the performativity of ethics: Implications
for communication governance. Big Data & Society, 7(1), 1–12. https://doi.org/10.1177/2053951720915939
Kind, C. (2020). The term ‘ethical AI’ is finally starting to mean something. VentureBeat. https://venturebeat.com/2020/08/23/
the-term-ethical-ai-is-finally-starting-to-mean-something/
Kukutai, T. & Taylor, J. (Eds.). (2016). Indigenous data sovereignty. ANU Press. https://doi.org/10.22459/CAEPR38.11.2016
Kusner, M. J., & Loftus, J. R. (2020). The long road to fairer algorithms. Nature, 578(7793), 34–36. https://doi.org/10.1038/
d41586-020-00274-3
Kwet, M. (2019). Digital colonialism: US empire and the new imperialism in the Global South. Race & Class, 60(4), 3–26.
https://doi.org/10.1177/0306396818823172
Lee, T. B. (2021). Why Big Tech is facing regulatory threats from Australia to Arizona. Ars Technica. https://arstechnica.com/
tech-policy/2021/03/why-big-tech-is-facing-regulatory-threats-from-australia-to-arizona/
Lempert, R. (2019). Bezos World or Levelers: Can we choose our scenario? https://www.rand.org/pubs/external_publica-
tions/EP67822.html
Lepri, B., Oliver, N., Letouzé, E., Pentland, A., & Vinck, P. (2018). Fair, transparent, and accountable algorithmic decision-making
processes. Philosophy & Technology, 31(4), 611–627. https://doi.org/10.1007/s13347-017-0279-x
Levy, K., Chasalow, K., & Riley, S. (2021). Algorithms and decision-making in the public sector. Annual Review of Law and Social
Science, 17, 309–334. https://doi.org/10.1146/annurev-lawsocsci-041221-023808
Lewis, J. E., Abdilla, A., Arista, N., Baker, K., Benesiinaabandan, S., Brown, M., Cheung, M., Coleman, M., Cordes, A., Davison, J.,
Duncan, K., Garzon, S., Harrell, D. F., Jones, P.-L., Kealiikanakaoleohaililani, K., Kelleher, M., Kite, S., Lagon, O., Leigh, J., …
Whaanga, H. (2020). Indigenous protocol and artificial intelligence position paper. Indigenous protocol and artificial intelli-
gence working group and the Canadian institute for advanced research. https://spectrum.library.concordia.ca/986506/
Lucassen, L. (2010). A brave new world: The left, social engineering, and eugenics in twentieth-century Europe. International
Review of Social History, 55(2), 265–296. https://doi.org/10.1017/S0020859010000209
Lury, C. (2018). Introduction: Activating the present of interdisciplinary methods. In C. Lury, R. Fensham, A. Heller-Nicholas,
S. Lammes, A. Last, M. Michael & E. Uprichard (Eds.), Routledge handbook of interdisciplinary research methods (pp. 1–25).
Routledge. https://doi.org/10.4324/9781315714523
Lyon, D. (2003). In Surveillance as social sorting: Privacy, risk, and digital discrimination. Routledge.
Mannheim, K. (1940). Man and society in an age of reconstruction: Studies in modern social structure. Kegan Paul. (E. Shils,
Trans.).
Maris, E. (2022). The humanities can’t save big tech from itself. Wired. https://www.wired.com/story/ethicis-big-tech-humanities/
McGranahan, C. (2016). Theorizing refusal: An introduction. Cultural Anthropology, 31(3), 319–325.
Mehrabi, N., Morstatter, F., Saxena, N., Lerman, K., & Galstyan, A. (2019). A survey on bias and fairness in machine learning.
ArXiv:1908.09635 [Cs]. http://arxiv.org/abs/1908.09635
Milan, S., & Treré, E. (2019). Big data from the south(s): Beyond data universalism. Television & New Media, 20(4), 319–335.
https://doi.org/10.1177/1527476419837739
Mitchell, S., Potash, E., Barocas, S., D’Amour, A., & Lum, K. (2020). Prediction-based decisions and fairness: A catalogue of choic-
es, assumptions, and definitions. ArXiv:1811.07867 [Stat]. http://arxiv.org/abs/1811.07867
ZAJKO 15 of 16
Moats, D., & Seaver, N. (2019). “You social scientists love mind games”: Experimenting in the “divide” between data science
and critical algorithm studies. Big Data & Society, 6(1), 1–11. https://doi.org/10.1177/2053951719833404
Mohamed, S., Png, M.-T., & Isaac, W. (2020). Decolonial AI: Decolonial theory as sociotechnical foresight in artificial intelli-
gence. Philosophy & Technology, 33(4), 659–684. https://doi.org/10.1007/s13347-020-00405-8
Newlands, G. (2021). Lifting the curtain: Strategic visibility of human labour in AI-as-a-Service. Big Data & Society, 8(1), 1–14.
https://doi.org/10.1177/20539517211016026
Noble, S. U. (2018). Algorithms of oppression: How search engines reinforce racism. NYU Press.
O’Neil, C. (2016). Weapons of math destruction: How big data increases inequality and threatens democracy. Crown.
Ossewaarde, M., & Gulenc, E. (2020). National varieties of artificial intelligence discourses: Myth, utopianism, and solution-
ism in West European policy expectations. Computer, 53(11), 53–61. https://doi.org/10.1109/MC.2020.2992290
Park, S., & Humphry, J. (2019). Exclusion by design: Intersections of social, digital and data exclusion. Information, Communi-
cation & Society, 22(7), 934–953. https://doi.org/10.1080/1369118X.2019.1606266
Pascale, C.-M. (Ed.). (2013). Social inequality & the politics of representation. SAGE.
Perlstadt, H. (2007). Applied sociology. In C. D. Bryant & D. L. Peck (Eds.), 21st century sociology (pp. 342–352). SAGE.
Podgórecki, A., Alexander, J., & Shields, R. (Eds.). (1996). Social engineering. Carleton University Press.
PWGSC. (2018). Artificial Intelligence (AI) pilot project for surveillance of suicide-related behaviours using social media
(1000203413). https://buyandsell.gc.ca/procurement-data/tender-notice/PW-18-00840702
Reevely, D. (2021). Federal rules on AI too narrow and risk ‘damaging public trust’: Internal review. The Logic. https://thelogic.co/
news/federal-rules-on-ai-too-narrow-and-risk-damaging-public-trust-internal-review/
Roberge, J., & Castelle, M. (2021). Toward an end-to-end sociology of 21st-century machine learning. In J. Roberge & M.
Castelle (Eds.), The cultural life of machine learning: An incursion into critical AI studies (pp. 1–29). Springer International
Publishing.
Roberts, D. E. (2019). Digitizing the carceral state. Harvard Law Review, 132(6), 1695–1728.
Roberts, D. E., & Rollins, O. (2020). Why sociology matters to race and biosocial science. Annual Review of Sociology, 46,
195–214.
Sampath, P. G. (2021). Governing artificial intelligence in an age of inequality. Global Policy, 12(S6), 21–31. https://doi.
org/10.1111/1758-5899.12940
Scheuerman, M. K., Pape, M., & Hanna, A. (2021). Auto-essentialization: Gender in automated facial analysis as extended
colonial project. Big Data & Society, 8(2), 1–15. https://doi.org/10.1177/20539517211053712
Schou, J., & Pors, A. S. (2019). Digital by default? A qualitative study of exclusion in digitalised welfare. Social Policy and Ad-
ministration, 53(3), 464–477. https://doi.org/10.1111/spol.12470
Schuilenburg, M. & Peeters, R. (Eds.). (2021). The algorithmic society: Technology, power, and knowledge. Routledge.
Science Academies, G. 7 (2019). Artificial intelligence and society. https://rscsrc.ca/sites/default/files/Artificial%20intelli-
gence%20and%20society%20G7%202019.pdf
Selwyn, N. (2004). Reconsidering political and popular understandings of the digital divide. New Media & Society, 6(3), 341–
362. https://doi.org/10.1177/1461444804042519
Seyfert, R. (2021). Algorithms as regulatory objects. Information, Communication & Society, 0(0), 1–17. https://doi.org/10.10
80/1369118X.2021.1874035
Shah, A. (2018). AI for the good of society. BBC Blogs. https://www.bbc.co.uk/blogs/internet/entries/7e49f841-85af-
4455-a8b0-c16e6279176c
Shestakofsky, B. (2017). Working algorithms: Software automation and the future of work. Work and Occupations, 44(4),
376–423. https://doi.org/10.1177/0730888417726119
Shestakofsky, B. (2020). Stepping back to move forward: Centering capital in discussions of technology and the future of
work. Communication and the Public, 5(3–4), 129–133. https://doi.org/10.1177/2057047320959854
Shestakofsky, B., & Kelkar, S. (2020). Making platforms work: Relationship labor and the management of publics. Theory and
Society, 49(5), 863–896. https://doi.org/10.1007/s11186-020-09407-z
Silberg, J., & Manyika, J. (2019). Notes from the AI frontier: Tackling bias in AI (and in humans). https://www.mckinsey.com/~/
media/mckinsey/featured insights/artificial intelligence/tackling bias in artificial intelligence and in humans/mgi-tack-
ling-bias-in-ai-june-2019.ashx
Simonite, T. (2021). What really happened when google ousted Timnit Gebru. Wired. https://www.wired.com/story/google-
timnit-gebru-ai-what-really-happened
Simpson, A. (2014). Mohawk interruptus: Political life across the borders of settler states. Duke University Press.
Sinanan, J., & McNamara, T. (2021). Great AI divides? Automated decision-making technologies and dreams of development.
Continuum, 35, 1–760. https://doi.org/10.1080/10304312.2021.1983257
Singh, S., & Steeves, V. (2020). The contested meanings of race and ethnicity in medical research: A case study of the
DynaMed point of care tool. Social Science & Medicine, 265, 113112. https://doi.org/10.1016/j.socscimed.2020.113112
Solon, O. (2020). While Facebook works to create an oversight board, industry experts formed their own. NBC News. https://
www.nbcnews.com/tech/tech-news/facebook-real-oversight-board-n1240958
16 of 16 ZAJKO
Stark, L., Greene, D., & Hoffmann, A. L. (2021). Critical perspectives on governance mechanisms for AI/ML systems. In J.
Roberge & M. Castelle (Eds.), The cultural life of machine learning: An incursion into critical AI studies (pp. 257–280).
Springer International Publishing.
Stark, L., & Hutson, J. (forthcoming). Physiognomic artificial intelligence. Fordham Intellectual Property, Media and Entertain-
ment Law Journal. https://doi.org/10.2139/ssrn.3927300
Stinson, C. (2021). The dark past of algorithms that associate appearance and criminality. American Scientist, 109(1), 26–29.
Suresh, H., & Guttag, J. V. (2020). A framework for understanding unintended consequences of machine learning.
ArXiv:1901.10002 [Cs, Stat]. http://arxiv.org/abs/1901.10002
Susskind, J. (2018). Future politics: Living together in a world transformed by tech. Oxford University Press.
Turner, J. (2019). Robot rules: Regulating artificial intelligence. Palgrave Macmillan.
Turner, J. H. (2001). Social engineering: Is this really as bad as it sounds? Sociological Practice, 3(2), 99–120.
Tusikov, N. (2017). Chokepoints: Global private regulation on the internet. University of California Press.
van Dijk, J. A. G. M. (2005). The deepening divide: Inequality in the information society. SAGE.
van Dijk, J. A. G. M. (2006). Digital divide research, achievements and shortcomings. Poetics, 34(4–5), 221–235. https://doi.
org/10.1016/j.poetic.2006.05.004
Vicsek, L. (2020). Artificial intelligence and the future of work – lessons from the sociology of expectations. International
Journal of Sociology & Social Policy, 41(7/8), 842–861. https://doi.org/10.1108/IJSSP-05-2020-0174
Wajcman, J. (2017). Automation: Is it really different this time? British Journal of Sociology, 68(1), 119–127. https://doi.
org/10.1111/1468-4446.12239
Walch, K. (2020). Why the race for AI dominance is more global than you think. Forbes. https://www.forbes.com/sites/
cognitiveworld/2020/02/09/why-the-race-for-ai-dominance-is-more-global-than-you-think/
Walter, M., Kukutai, T., Carroll, S. R., & Rodriguez-Lonebear, D. (Eds.). (2020). Indigenous data sovereignty and policy. Rout-
ledge. https://doi.org/10.4324/9780429273957
Ward, L. F. (1906). Applied sociology: A treatise on the conscious improvement of society by society. Ginn & Company.
Watts, D. J. (2017). Should social science be more solution-oriented? Nature Human Behaviour, 1(1), 1–5. https://doi.
org/10.1038/s41562-016-0015
Weinstein, J. (2000). The place of theory in applied sociology: A reflection. Theory & Science, 1(1). https://theoryandscience.
icaap.org/content/vol001.001/01weinstein_revised.html
Wellman, B. & Haythornthwaite, C. (Eds.). (2002). The internet in everyday life. Blackwell.
West, K. G., Schechner, S., McMillan, R., John. (2019). How Google interferes with its search algorithms and changes your
results. Wall Street Journal. https://www.wsj.com/articles/how-google-interferes-with-its-search-algorithms-and-
changes-your-results-11573823753
Williamson, B., & Eynon, R. (2020). Historical threads, missing links, and future directions in AI in education. Learning, Media
and Technology, 45(3), 223–235. https://doi.org/10.1080/17439884.2020.1798995
Winner, L. (1980). Do artifacts have politics? Dædalus, 109(1), 121–136.
Zajko, M. (2021). Conservative AI and social inequality: Conceptualizing alternatives to bias through social theory. AI & Society,
36(3), 1047–1056. https://doi.org/10.1007/s00146-021-01153-9
Zarkadakis, G. (2020). Cyber republic: Reinventing democracy in the age of intelligent machines. MIT Press.
Zuboff, S. (2018). The age of surveillance capitalism: The fight for a human future at the new frontier of power. PublicAffairs.
AUT HOR B I OGRAPHY
Mike Zajko is an Assistant Professor in the Department of History and Sociology, University of British Columbia,
Okanagan Campus. He holds a PhD in Sociology from the University of Alberta. His current research examines
automated decision-making in the public sector and the use of machine-readable rules to govern behaviour.
How to cite this article: Zajko, M. (2022). Artificial intelligence, algorithms, and social inequality:
Sociological contributions to contemporary debates. Sociology Compass, 16(3), e12962. https://doi.
org/10.1111/soc4.12962

ZAJKO - Artificial Intelligence Algorithms and Social Inequality Sociological Contributions To Contemporary Debates

Uploaded by

Copyright:

Available Formats

ZAJKO - Artificial Intelligence Algorithms and Social Inequality Sociological Contributions To Contemporary Debates

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

ZAJKO - Artificial Intelligence Algorithms and Social Inequality Sociological Contributions To Contemporary Debates

Uploaded by

Copyright:

Available Formats

Received: 19 June 2021 Revised: 19 January 2022 Accepted: 19 January 2022

Artificial intelligence, algorithms, and social

Department of History and Sociology, The

Sociology Compass. 2022;16:e12962. wileyonlinelibrary.com/journal/soc4 1 of 16

limitations of utilizing these new technologies to address so-

3 | BIAS AS SOCIETY'S “UNEQUAL GROUND TRUTH”

4 | SOCIAL INEQUALITY AND TECHNOLOGY: THE VIEW FROM SOCIOLOGY

5 | THE FUTURE OF INEQUALITY AND SOCIOLOGY'S RESPONSE

5.2 | Fighting inequality through technology

5.2.1 | Interdisciplinary social engineering

5.3 | Governance of algorithmic governance

AUT HOR B I OGRAPHY

You might also like