Oct 10, 2025
Via our Friction for Accountability in Conversational Transactions (FACT) AI Exploration, we are actively engaging the research community in addressing the challenges of accountable AI.
As part of this effort, a team of researchers from Colorado State University, University of Illinois Urbana-Champaign, Brandeis University, Tufts University, University of Southern California, and Northeastern University—performers on FACT—will convene the first workshop on Optimal Reliance and Accountability in Interactions with Generative Language Models (ORIGen) at the second annual Conference on Language Modeling (COLM).
This conference showcases advances in the field of natural language processing (NLP).
With the rapid integration of generative AI—particularly large language models (LLMs)—into personal, educational, business, and governmental workflows, these systems are increasingly being treated as collaborators in human-AI teams. In such contexts, insufficient reliance on AI may undermine potential gains, while conversely, over-trusting AI systems may be catastrophic.
Addressing this tension by establishing optimal levels of reliance within interactive human-AI frameworks is a pressing challenge in the era of rapidly advancing language models.
The ORIGen workshop aims to create a new interdisciplinary forum for exploring these questions. We invite perspectives from AI, NLP, human-computer interaction, cognitive science, psychology, and education to examine how to mediate human-LLM interactions, mitigate overreliance, and foster accountability in collaborative decision-making.
The workshop will take place on Oct. 10, 2025, in Montreal. Visit the workshop website for additional information.
