Project

LLM Targeted Underperformance Disproportionately Impacts Vulnerable Users

Elinor Poole-Dayan

Project Contact:

Groups

This research, recently presented at the AAAI 2026 conference in Singapore by Elinor Poole-Dayan, finds that leading AI models perform worse for users with lower English proficiency, less formal education, and non-U.S. origins.

See the full publication: https://arxiv.org/abs/2406.17737

Research Topics

#human-computer interaction #artificial intelligence

Project at a glance

Publication Research

LLM Targeted Underperformance Disproportionately Impacts Vulnerable Users

Poole-Dayan, Elinor, Deb Roy, and Jad Kabbara. "LLM Targeted Underperformance Disproportionately Impacts Vulnerable Users." Proceedings of the AAAI Conference on Artificial Intelligence (Jan. 2026). https://arxiv.org/abs/2406.17737.

Past Member

Person People

Elinor Poole-Dayan

Former Graduate Student

Person People

Deb Roy

Professor of Media Arts and Sciences; Toshiba Professor of Media Arts and Sciences

Past Member

Person People

Jad Kabbara

Former Research Scientist