In our next week’s Perspectives on Science Seminar, Inkeri Koskinen and Santtu Kauppila (TINT / Helsinki) will give a talk titled “Towards a Satisfactory Social Epistemology of AI-based Science: Forward-looking trust in collective pursuit”.

The seminar will take place in Metsätalo room 10, and online on Zoomfrom 14:15 to 15:45 on Monday, March 2, 2026.

Perspectives on Science is a research seminar that brings together experts from the philosophy of science and several fields of science studies. It is organized by TINT – Centre for Philosophy of Social Science at the University of Helsinki. More information about the seminar can be found on the TINT webpage https://tint.helsinki.fi.

What: “Towards a Satisfactory Social Epistemology of AI-based Science: Forward-looking trust in collective pursuit” by Inkeri Koskinen and Santtu Kauppila

When: Monday 02.03.2026, from 2 to 4 pm (EEST, Helsinki time).

Where: Metsätalo room 10 and Zoom. 

Zoom link: Contact mirja-leena.zgurskaya@helsinki.fi for the Zoom link.

Abstract:

“Towards a Satisfactory Social Epistemology of AI-based Science: Forward-looking trust in collective pursuit”

Understanding the epistemic role of forward-looking trust in the collective pursuit of scientific ideas, theories, and research programmes helps in the formulation of a satisfactory social epistemology of AI-based science. In this paper we spell out this claim and defend it. To illustrate our argument, we present a large case study based on a literature review.

We focus on the kind of AI-based science that has already taken distinct forms, meaning that there are established research practices that one can study. Consequently we do not discuss the use of generative AI in science. Our argument, nevertheless, pertains to all science, not just AI-based science. It should therefore be also relevant in research based on the use of generative AI.

Our main argument can be summarised as follows. In the social epistemology of science, there is a widely shared understanding of the necessity of relationships of trust in the kind of collective knowledge production that is characteristic of contemporary science. As John Hardwig (1985, 1991, p. 702) has argued, if we want to claim that individuals taking part in collective knowledge production know the collectively produced results, we must accept that “trust in the moral sense” has a role in the justification of the individual’s beliefs. And as Torsten Wilholt (2013) has argued, if we accept that values have legitimate and ineliminable roles in all stages of research, then individuals taking part in collective production must to some degree trust their colleagues, not just rely on them. The claim that trust in a relatively thick, moral sense has an ineliminable role in collective knowledge production is, as noted, generally accepted in the social epistemology of science. This view, however, does not seem compatible with contemporary AI-based science, where researchers rely on epistemically opaque AI tools that are not moral agents and therefore not suitable candidates for moral trust (Koskinen, 2024a).

We argue that the arguments presented by Hardwig, Wilholt, and many other philosophers taking part in the discussion about trust in science do not cover all important roles that relationsips of trust play in collective scientific knowledge production. This is because the discussion does not consider other cognitive attitudes than belief. As a result, it focuses on what can be called backward-looking trust: trust in someone having already done something. When we take into account many other cognitive attitudes that are important in science – for instance, conditional acceptance and pursuit – we notice that also forward-looking trust has an ineliminable epistemic role in collective scientific knowledge production. We argue that such trust is trust in a relatively thick sense, and that the trustees are future members of the research community  – that is, full epistemic and moral agents, not epistemically opaque tools. We further argue, and demonstrate with a case study, that when pay attention to the role that forward-looking trust plays in the justification of collective pursuit, we are much closer to having a satisfactory social epistemology of AI-based science.

Bio:

Inkeri Koskinen is a philosopher of science working as an Academy of Finland Research Fellow in Practical philosophy, University of Helsinki, and a member of the Centre for Philosophy of Social Science (TINT). Santtu Kauppila is a master’s student in philosophy and Koskinen’s research assistant.

____

If you have any questions about the seminar, do not hesitate to contact mirja-leena.zgurskaya@helsinki.fi.