Price
Free
An event organized by the IVADO Regroupement – Ethics, EDI, and Indigenous Engagement
The widespread use of artificial intelligence (AI) systems is transforming the way we describe the actions of machines, which are now referred to as multimodal, autonomous, conversational, decision-making “agents,” etc.
However, the concept of agency is plural and contextual, and there seems to be a real tension between the computational use of the term “agent” or “agentic” and its philosophical use. AI models often describe agents without necessarily assuming subjectivity or strong moral responsibility, while philosophy questions the agent as a member of a moral and political community. This divergence encourages conceptual shifts, because in public (and sometimes scientific) debate, these registers are mixed by attributing “decisions,” “choices,” “behaviors,” and even “intentions” to AI. It is therefore urgent to clarify the uses of the term “agent,” the notion of “artificial agency,” and agentic AI.
This conference offers a structured space for interdisciplinary dialogue between researchers in philosophy and artificial intelligence. It aims to strengthen the theoretical intelligibility of discourse on “AI agents” while fostering critical reflection on the social, political, and normative implications of artificial agency.
Schedule
8:30 a.m. – 9:00 a.m. : Welcome and Registration
9:00 am – 9:30 a.m. : Welcome address
Aaron Courville (Université de Montréal)
9:30 a.m. – 11:00 a.m. : What are the minimal conditions (necessary and/or sufficient) for speaking of artificial agency from both philosophical and technical perspectives? Which ones truly matter? How do computer scientists define an agent?
Jocelyn Maclure (McGill University), Hugo C. Lefebvre (McGill University), Clemence Bergerot (Humboldt University), Jeff Sebo (NYU)
11:00 a.m. – 11:15 a.m. : Break
11:15 a.m. – 11:45 a.m. : Is consciousness or phenomenological experience necessary, or are behavioral properties sufficient for agency ? In other words, how can we compare philosophical criteria and technical criteria without talking about different things under the same concept?
Joé Martineau (HEC Montréal), Patricia Gautrin (Université de Montréal), Alexis Morin-Martel (McGill University)
12:45 p.m. – 2:00 p.m. : Lunch
2:00 p.m. – 3:30 p.m. : What shared vocabulary could help avoid anthropomorphic slippage? Can concepts from the philosophy of action (intentionality, deliberation, acting for reasons, etc.) be translated into frameworks such as BDI (Belief–Desire–Intention) (Rao & Georgeff 1991, 1998), decision theory, or LLM agents?
Annie Pullen Sansfaçon (Université de Montréal), Audrey Durand (Université Laval), Dominic Martin (Université du Québec à Montréal), Anandi Hattiangadi (Stockholm University)
3:30 p.m. – 3:45 p.m. : Break
3:45 p.m. – 5:15 p.m. : Under what conditions can an artificial system be legitimately considered an agent: is externally observed behavioral adequacy sufficient, or does attributing genuine agency require access to its internal states and constitutive mechanisms? What about collective agents?
Daniel Weinstock (McGill University), Xabier Barandiaran (University of the Basque Country), Tegan Maharaj (HEC Montréal), Marion Korosec-Serfaty (Université du Québec à Montréal)
5:15 p.m. – 5:30 p.m. : Closing remarks
Daniel Weinstock (McGill University)
*Program may be subject to changes.
