Bio

Aadesh Salecha is a computer scientist who works on integrating Artificial Intelligence and Psychology. He graduated from the University of Minnesota, where he worked with Prof. Jaideep Srivastava on misinformation spread and its mitigation mechanisms. At Stanford, he works with Prof. Johannes Eichsteadt on using cognitive psychology and psychometrics to understand bias development in Large Language Models. He is also collaborating the effort on the creation of new-age psychological interventions using AI to democratize access and improve personalization and retention. His work is focused on using computational methods for societal good by facilitating the measurement and causal analysis of population health metrics like drug abuse, subjective-wellbeing, etc.

Contact

Additional Info

  • Mail Code: 9025

All Publications

  • Large language models display human-like social desirability biases in Big Five personality surveys. PNAS nexus Salecha, A., Ireland, M. E., Subrahmanya, S., Sedoc, J., Ungar, L. H., Eichstaedt, J. C. 2024; 3 (12): pgae533

    Abstract

    Large language models (LLMs) are becoming more widely used to simulate human participants and so understanding their biases is important. We developed an experimental framework using Big Five personality surveys and uncovered a previously undetected social desirability bias in a wide range of LLMs. By systematically varying the number of questions LLMs were exposed to, we demonstrate their ability to infer when they are being evaluated. When personality evaluation is inferred, LLMs skew their scores towards the desirable ends of trait dimensions (i.e. increased extraversion, decreased neuroticism, etc.). This bias exists in all tested models, including GPT-4/3.5, Claude 3, Llama 3, and PaLM-2. Bias levels appear to increase in more recent models, with GPT-4's survey responses changing by 1.20 (human) SD and Llama 3's by 0.98 SD, which are very large effects. This bias remains after question order randomization and paraphrasing. Reverse coding the questions decreases bias levels but does not eliminate them, suggesting that this effect cannot be attributed to acquiescence bias. Our findings reveal an emergent social desirability bias and suggest constraints on profiling LLMs with psychometric tests and on this use of LLMs as proxies for human participants.

    View details for DOI 10.1093/pnasnexus/pgae533

    View details for PubMedID 39691446

  • Influence of emotions on coping behaviors in crisis: a computational analysis of the COVID-19 outbreak JOURNAL OF COMPUTATIONAL SOCIAL SCIENCE Xu, H., Sudheendra, S., Huh, J., Salecha, A., Srivastava, J. 2024

    View details for DOI 10.1007/s42001-024-00282-7

    View details for Web of Science ID 001215061500001

  • The Language of Conflict Transformation: Assessing Psychological Change Patterns in Israeli-Palestinian Track Two Interactive Problem Solving NEGOTIATION AND CONFLICT MANAGEMENT RESEARCH Fink, O., Graf, W., Subrahmanya, S., Salecha, A., Eichstaedt, J. C. 2024; 17 (2): 130-152

    View details for DOI 10.34891/svxv-s665

    View details for Web of Science ID 001247374700002