Stephen Casper
Harvard Kennedy School assistant professor working on AI safeguards, incidents, and governance
- Assistant Professor; Faculty Affiliate with the Harvard School of Engineering and Applied Sciences, Harvard University (Harvard Kennedy School)1
Stephen Casper is a computer scientist and an Assistant Professor at the Harvard Kennedy School, as well as a Faculty Affiliate with the Harvard School of Engineering and Applied Sciences. Casper works on AI safeguards, incidents, and governance, and completed a PhD at MIT. Casper is first author of a survey of open problems and fundamental limitations of reinforcement learning from human feedback (RLHF), and of a paper examining the limitations of black-box AI audits and the advantages of white-box and outside-the-box access.123
Work
Open Problems and Fundamental Limitations of Reinforcement Learning from Human Feedback (external site: arxiv.org)Published research
First author of the 2023 paper, which surveys open problems and limitations of RLHF, reviews techniques to understand, improve, and complement it, and proposes auditing and disclosure standards for RLHF systems.21
Black-Box Access is Insufficient for Rigorous AI Audits (external site: arxiv.org)Published research
First author of the 2024 paper examining the limitations of black-box audits of AI systems and the advantages of white-box and outside-the-box access for auditors.31
This profile covers public professional work only, from the sources below. It is not an endorsement or a ranking. To correct or remove it, use the correction form.