Seminar Agency between Biology and Philosophy 2025/2026

Here, you will find the 2025-2026 programme, and general information about our seminar.

The introductory post can be found here.

Next session

13 May, 4 to 5 UK time

Session led by Leonard Dung

Title: Distinguishing AI roleplay and agency: The case of deception

Abstract: Large language model based systems display apparently sophisticated and human-like behavior. Here, we explore the view that this behavioral competence can be explained entirely by the model’s capacity to roleplay various actual and fictional human agents. This view promises to reconcile inflationary views of LLM mentality that explain their behavior in recourse to mental states with deflationary views by denying that the LLMs themselves, rather than the agents they simulate, have mental states. We first explain the different versions of the roleplay view, distinguishing their main commitments. Then, we argue that, without dedicated empirical testing, we cannot decide whether LLM behavior is "roleplay all the way down" (Shanahan et al. 2023) or whether some LLM behavior should be explained without recourse to roleplay. However, focusing on the case of AI deception research, we show problems in empirically testing roleplay views while arguing that it matters for AI safety whether LLMs are merely roleplaying or something more. Finally, we propose an account of how future empirical research could in principle confirm or disconfirm roleplay views.

Join Zoom Meeting:

https://ucl.zoom.us/j/94845332305

Meeting ID: 948 4533 2305

Programme for 2025/2026

This year, we will meet each second Wednesday of each month, from 4pm to 5pm (UK time).

Upcoming dates: 

• 12 November: Session on the minimal viable mechanisms of biological agency (session led by Daria Zakharova). (Report available on the Google Drive)*

• 10 December: "A good life on their own?" Session led by Angie Pepper on how to conceive of animal interests in freedom. (Report available on the Google Drive)*

• 14 January : Session on how to promote agency for captive animals, with a focus on elephants (session led by Lisa Yon) (Report available on the Google Drive)*

• 11 February: Session on play as a problematic case of agency (session led by Marek Špinka) (Report available on the Google Drive)*

• 25 March: Session on agency and cognitive capacities led by Hugh Desmond (Report available on the Google Drive)*

• 8 April: Ecological Agency: A metaphysics for Dialectical Biology (session led by Denis Walsh) (Report available on the Google Drive)*

• 13 May: Session on AI agency "Distinguishing AI roleplay and agency: The case of deception" (session led by Leonard Dung)

• 10 June: "Normative agency beyond humans?" Session led by Christopher Joseph Gonzalez An)

*Abstracts, slides, references, and reports of each session can be found in our share Google Drive, when available. Please contact mathilde.tahar@ucl.ac.uk for more information on how to join the mailing list, and access the Google Drive.

Join Zoom Meeting:

https://ucl.zoom.us/j/94845332305

Meeting ID: 948 4533 2305

Session Format

Each session will be structured as follows:

1. A 15–20-minute presentation, introducing a conceptual or empirical challenge related to agency.

2. A 40–45-minute discussion, aimed at collaboratively exploring the topic.

Topics

The following topics will guide our discussions throughout the series (not necessarily in this order):

1. Autonomy

   • How do we define autonomy?

   • Does it imply moral obligations? Are agents moral agents and/or moral recipients?

2. Choice, Decision-Making, and Intentionality

   • What does it mean for a nonhuman animal to have a choice or the ability to make decisions?

   • Is choice necessarily linked to intentionality?

   • How can choice or intentionality be recognised through behaviours?

   • How can we integrate these ideas into methods and experiments?

   • How can we avoid anthropomorphism?

3. Niche Construction and Effects on Ecology/Evolution

   • What makes a living system an “evolutionary” or “ecological” agent?

   • Are all living systems or organisms agents?

4. Levels and Degrees/Forms of Agency

   • Are there multiple levels of agency?

   • Are there degrees of agency

   • Are there distinct forms of agency?

5. Studying Agency Without Manipulating It

   • How can we study agency ethically, and when is manipulation/human interference justified or not?

6. Agency and Animal Welfare

   • How can agency considerations be integrated into welfare practices?

7. The Conceptual Role of Agency

   • What role does the concept of agency play in research?

Join Us

If you’re interested in joining the mailing list, accessing our shared Google Drive folder, or learning more about the seminar, please email: mathilde.tahar@ucl.ac.uk.