Sub-theme 11: [SWG] Explaining AI in the Context of Organizations
Call for Papers
“The computer says no” is a statement most are not happy to accept. Yet as algorithmic decision-making, increasingly reliant
on machine learning ((Faraj, Pachidi, & Sayegh, 2018), becomes a normal component of organizational processes, employees,
managers, and clients are facing a reality where machines make ‘decisions’ that are implemented without meaningful avenues
for questioning and redress.
As users and stakeholders working with such AI systems grow increasingly aware
of issues of bias, discrimination and inaccuracy (AI NOW, 2018) and the public concerns about a black-box society (Pasquale,
2015) and surveillance capitalism (Zuboff, 2019) increase, the need to ensure that systems are fair, transparent and accountable
have rapidly grown. The intense focus on these issues has brought forward a general consensus that “explainability” forms
a critical and urgent move towards ensuring that AI systems can be trusted and safely integrated throughout many sensitive
and high-impact domains (HLEG AI, 2019).
However, the discussion and developments around Explainable AI have
thus far been held in relatively siloed communities of technologists, and policy makers and ethicists, exposing a gap that
may prevent meaningful and productive progress: the lack of integration between policy makers, technologists and the contexts
of practice aggravates what is called the “sociotechnical” gap between technical feasibility and social requirements
(Ackerman, 2000). This phenomenon has been identified as the main hurdle in ensuring the adoption and economic success of
new technologies and media.
This gap arises in part because technologists have mostly developed technical
techniques to “interpret” the internal workings of algorithms and machine learning models (Došilović, Brčić, & Hlupić,
2018; Guidotti et al., 2018). While such techniques are desperately needed, it remains a challenge to bridge the gap between
these technical interpretations and the actual social needs of a context of use and organizational practice, for
example, with regards to how such tools are embedded in specific knowledge practices (Andrejevic, 2020).
In
particular, what is lacking in the current discussions on Explainable AI is empirical evidence from organizations that employ
AI tools. As long as we do not incorporate organizational insights about the need for Explainable AI and existing local solutions,
many of the present and future technological solutions and political guidelines might have missed the point. As Introna (2016)
points out, hard-to-scrutinize algorithms are “subsumed in the flow of daily practices” and this urges us to study and develop
explainability that is situated in the web of different domains of knowledge and subjectivities that are enacted through governing
practices. To understand both how explanations around AI play out in practice, and in order to close the gap between technical
solutions and social requirements, we need to expand the view of Explainable AI to include an organizational practice perspective.
An organizational practice perspective on Explainable AI invites us to look at how AI is used and developed
in situ (Hafermalz & Huysman, 2021). In particular, we want to stress that the development and use of AI relies on a host
of stakeholders: for example data scientists, domain experts, end-users, brokers, managers, and more (Waardenburg, Huysman,
& Agterberg, 2021; Zhang et al., 2020). “Explanations” may therefore play a role beyond that of aspiring to technical
transparency, for example in matters of persuasion, politics, and rhetoric (Alvesson, 2001). Further, explanations that pertain
to AI may be constructed in ways that range from mostly social (Runde & de Rond, 2010) to mostly technical; with intentions
that range from ethical transparency to influence. Unearthing the specificities of sociomaterial practices of explanation
across different organizational contexts will add needed richness and nuance to the broader Explainable AI (XAI) initiative.
An organizational practice perspective is suited to answering the following questions, that until so far are
left mostly unexplored:
What does Explainable AI mean in the context of organizations, and what are its (unintended) consequences?
Who is the actor or actors in need of Explainable AI and do these various actors call for different solutions (Kirsch, 2017)?
When do these actors need explanations from AI? Why do we need Explainable AI in the context of organizational practice?
Addressing
the above questions will give us better insight into how to design Explainable AI that addresses the sociotechnical
gap between design and use (Bailey & Barley, 2020).
Contributions can include but are not
limited to:
Case studies of organizational uses of AI with attention paid to explainability, accountability, visibility, and transparency (Ananny & Crawford, 2018; Flyverbom et al., 2016).
Studies of Explainable AI in terms of knowing practices (Pachidi et al., 2020), knowledge management, the shareability and automation of knowledge (Andrejevic, 2020).
Explorations of open AI initiatives (such as OpenAI) and their organizational principles.
Conceptual reflections on the limits of transparency and explainability, also in relation to organizational opacities (Burrell, 2016; Geiger, 2017; Roberts, 2009; Tsoukas, 1997).
Empirical process studies of how explainability is considered and/or 'built in' during the design and development of AI systems.
Methodological reflections on reverse engineering or otherwise opening up the ‘black boxes’ of AI (Kitchin, 2017), archaeologies of the operations of AI (Mackenzie, 2017).
Empirically informed ethical reflections on the use of AI in organizational contexts.
A stakeholder perspective on Explainable AI in organizations, that looks at explanations as rhetorical device rather than (only) an avenue for achieving transparency.
References
- Ackerman, M.S. (2000): “The intellectual challenge of CSCW: the gap between social requirements and technical feasibility.” Human–Computer Interaction, 15 (2–3), 179–203.
- AI NOW (2018): AI Now Report 2018. Retrieved from https://ainowinstitute.org/AI_Now_2018_Report.pdf.
- Alvesson, M. (2001): “Knowledge work: Ambiguity, image and identity.” Human relations, 54 (7), 863–886.
- Ananny, M., & Crawford, K. (2018): “Seeing without knowing: Limitations of the transparency ideal and its application to algorithmic accountability.” New Media & Society, 20 (3), 973–989.
- Andrejevic, M. (2020): “Data Civics: A Response to the ‘Ethical Turn’.” Television & New Media, 21 (6), 562–567.
- Bailey, D.E., & Barley, S.R. (2020): “Beyond design and use: How scholars should study intelligent technologies.” Information and Organization, 30 (2), 100286. https://doi.org/10.1016/j.infoandorg.2019.100286.
- Burrell, J. (2016): “How the machine ‘thinks’: Understanding opacity in machine learning algorithms.” Big Data & Society, 3 (1), https://doi.org/10.1177%2F2053951715622512.
- Došilović, F.K., Brčić, M., & Hlupić, N. (2018): Explainable artificial intelligence: A survey. Paper presented at the 2018 41st International Convention on Information and Communication Technology, Electronics and Microelectronics (MIPRO), Opatija (Croatia), May 21–25, 2018; https://ieeexplore.ieee.org/document/8400040.
- Faraj, S., Pachidi, S. & Sayegh, K. (2018): "Working and organizing in the age of the learning algorithm." Information and Organization, 28 (1), 62–70.
- Flyverbom, M., Leonardi, P., Stohl, C., & Stohl, M. (2016): “Digital Age| The Management of Visibilities in the Digital Age – Introduction.” International Journal of Communication, 10, 98–109.
- Geiger, R.S. (2017): “Beyond opening up the black box: Investigating the role of algorithmic systems in Wikipedian organizational culture.” Big Data & Society, 4 (2), https://doi.org/10.1177%2F2053951717730735.
- Guidotti, R., Monreale, A., Ruggieri, S., Turini, F., Giannotti, F., Pedreschi, D. (2018): “A survey of methods for explaining black box models.” ACM Computing Surveys (CSUR), 51 (5), 1–42.
- Hafermalz, E., & Huysman, M. (2021): “Please Explain: Key Questions for Explainable AI Research from an Organizational Eerspective.” Morals + Machines, 1 (2), 10–23.
- HLEG AI (2019): Ethics Guidelines for Trustworthy AI. Retrieved from European Commission, Brussels: https://ec.europa.eu/newsroom/dae/document.cfm?doc_id=60419.
- Introna, L.D. (2016): "Algorithms, governance, and governmentality: On governing academic writing." Science, Technology, & Human Values, 41 (1), 17–49.
- Kirsch, A. (2017): Explain to Whom? Putting the User in the Center of Explainable AI. Paper presented at the Proceedings of the First International Workshop on Comprehensibility and Explanation in AI and ML 2017 co-located with 16th International Conference of the Italian Association for Artificial Intelligence (AI* IA 2017).
- Kitchin, R. (2017): "Thinking critically about and researching algorithms." Information, Communication & Society, 20 (1), 14–29.
- Mackenzie, A. (2017): Machine Learners: Archaeology of a Data Practice. Cambridge: MIT Press.
- Pachidi, S., Berends, H., Faraj, S., & Huysman, M. (2020): “Make Way for the Algorithms: Symbolic Actions and Change in a Regime of Knowing.” Organization science, 32 (1), 18–41.
- Pasquale, F. (2015): The Black Box Society. Cambridge: Harvard University Press.
- Roberts, J. (2009): “No one is perfect: The limits of transparency and an ethic for ‘intelligent’ accountability.” Accounting, Organizations and Society, 34 (8), 957–970.
- Runde, J., & de Rond, M. (2010): “Evaluating causal explanations of specific events.” Organization Studies, 31 (4), 431–450.
- Tsoukas, H. (1997): "The tyranny of light: The temptations and the paradoxes of the information society." Futures, 29 (9), 827–843.
- Waardenburg, L., Huysman, M., & Agterberg, M. (2021): Managing AI Wisely: From Development to Organizational Change in Practice. Cheltenham: Edward Elgar Publishing.
- Zhang, Z., Nandhakumar, J., Hummel, J. & Waardenburg, L. (2020): “Addressing the key challenges of developing machine learning AI systems for knowledge-intensive work.” MIS Quarterly Executive, 19 (4), Article 5, https://aisel.aisnet.org/misqe/vol19/iss4/5.
- Zuboff, S. (2019): The Age of Surveillance capitalism: The Fight for a Human Future at the New Frontier of Power. London: Profile Books.