TSElosophers meeting on 19 May 2025
Participants: Albrecht Becker, Annika Hasselblad, Samu Kantola, Kari Lukka, Nicolas Balcom Raleigh, Marja Turunen
Reading
Bechky, B. A., & Davis, G. F. (2024). Resisting the Algorithmic Management of Science: Craft and Community After Generative AI. Administrative Science Quarterly, 00018392241304403. https://doi.org/10.1177/00018392241304403
Summary
In this essay, Bechky and Davis (2025) discuss their worries in the face of the emerging generative AI tools that are introduced into a context characterised by increasing algorithmic management of research and universities. Digitalisation has already enabled, on the one hand, a ‘Cambrian explosion’ in the number of journals and journal articles and, on the other hand, an infrastructure and culture of stressing the quantitative understanding of research output and quality, deepening the publish-or-perish culture and the ensuing dominance of instrumentalism. Generative AI amplifies, Bechky and Davis argue, and accelerates these developments: “generative AI threatens not just to manage our work by algorithms but to have algorithms do the work itself” (p. 3). This further undermines the two pillars of scholarship: craft in doing research and community as the constitutive context of research activities. Bechky and Davis discuss possible strategies to preserve, or bring back, scholarship, including revitalising scholarship as a craft in our everyday doings and specifically in educating early career researchers. They also emphasize upholding the ideal of scholarship in our communities, e.g., in hiring, evaluation, reviewing, and editorial activities.
Discussion
In the discussion, TSElosophers fundamentally agreed with the analysis of Bechky and Davis (2025). Their worries connected to many earlier discussions in different TSElosophers meetings. They also see the threat to scholarship increasing due to the danger of AI tools in the context of the publish-or-perish culture, driving out deep thinking and deep reading. Some notable statements argued that using AI tools “violates community” and leads to the feeling of “not owning my craft” as a researcher. However, some also argued that the essay undervalues positive aspects of AI, such as its potential to democratise publishing or relieve us of “dirty work”. We concluded, however, that the authors probably saw the well-known functional potential of AI as a matter of course and, therefore, left it without mention.
Probably not surprising, a part of the discussion centred around pressures of the publish-or-perish culture, lack of time, and the seductiveness of turning towards generative AI tools in this situation. Its statements’ seeming authoritativeness, constant availability, and eternal friendliness may be especially seductive. Another strand of the discussion on publish-or-perish was about Goodhart’s law. While some stated that it is important to reflect the conditions under which a performance measure turning target becomes dysfunctional, others saw it as an inherent dilemma to need measures that unavoidably become targets and thus dysfunctional.
When discussing the strategies of preserving craft, community, and scholarship in the face of generative AI, TSElosophers concluded that most of the suggestions were not surprising and wondered if they have the potential to fend off the threats. We more deeply discussed, first, the role of individual responsibilities. While it is clear that even a systemic issue in social systems can be changed through actions of individuals, in the case of academia, senior researchers must carry the larger load. Secondly, we discussed evaluation processes. In this context, one of us mentioned that they do not favour overly long publication lists; rather, they are a bit suspicious about scholarly quality if there are too many publications produced in a short time. For early career researchers, it would be helpful to hear such views from senior researchers personally. The same holds for the third strategy we discussed and which Bechky and Davis term “shadow CV”: senior academics sharing not only success stories but also the often tedious and dispiriting journeys towards such successes as well as failures. Again, early career researchers feel that this is helpful. One of us reported on a group of PhD students sharing issues and problems among themselves, but they think that “sharing across hierarchical levels” would be specifically helpful.