We measure overreliance of AI users and find that AI help boosts accuracy and efficiency, but users quickly become overreliant, leaving them vulnerable.
WildSEEK introduces real-world information-seeking queries and an evaluation framework for assessing the reliability, safety, and fairness of language-model responses.
A controlled study of how explicit and implicit demographic cues affect large language model scoring, feedback, and question answering in educational assessment.
SimBench provides a standardized benchmark for evaluating how faithfully large language models simulate human behavior across diverse social and behavioral tasks.
We investigate grounded language learning through real-world data, by modelling a teacher-learner dynamics through the natural interactions occurring between users and search engines.