BECAREFUL: Building Embodied Conversational Agent Reliability by Exerting Friction through Uncertain Language Project aims to enhance decision making mechanisms for conversational embodied AI agents by reducing user over-reliance to possible misinformation from AI systems (i.e., due to AI hallucinations, AI sycophancy or misunderstanding the user due to low-bandwidth or unreliable communication situations)
News
-
January 30, 2025New Pre-Print: Better Slow than Sorry: Introducing Positive Friction for Reliable Dialogue Systems
-
January 15, 2025New Pre-Print: Towards Preventing Overreliance on Task-Oriented Conversational AI Through Accountability Modeling
Research
A Taxonomy for Positive Friction
Better Slow than Sorry: Introducing Positive Friction for Reliable Dialogue Systems
Accountability Modeling
Towards Preventing Overreliance on Task-Oriented Conversational AI Through Accountability Modeling
Sycophancy in Language Model
Accounting for Sycophancy in Language Model Uncertainty Estimation
ReSpAct Harmonizing Reasoning, Speaking, and Acting
ReSpAct: Harmonizing Reasoning, Speaking, and Acting Towards Building Large Language Model-Based Conversational AI Agents
People
Faculty
Dilek Hakkani-Tür
Gokhan Tur
Malihe Alikhani
Jesse Thomason
Postdocs
Suvodip Dey
Graduate Students
Institutes
Sponsors
This project is supported by the U.S. Defense Advanced Research Projects Agency (DARPA) Friction for Accountability in Conversational Transactions (FACT) program