CHAI Internship
About
The CHAI Internship Program is a structured research mentorship program run by the Center for Human-Compatible AI (CHAI), a research center founded in 2016 at the University of California, Berkeley by Stuart Russell and colleagues. CHAI's mission is to reorient AI research toward provably beneficial systems, focusing on the problem of ensuring that advanced AI remains aligned with human values and intentions. The internship program has been running since at least 2017, with the first documented cohort completing work in 2017-2018. It is designed to bring talented students and early-career professionals into the AI safety research field, serving both as a pipeline for future PhD students and researchers and as a way for professionals to explore whether AI safety research aligns with their career goals. Interns work on individual research projects under the mentorship of a CHAI PhD student or postdoc, receiving guidance from conception through publication. Most projects result in first-author workshop papers at machine learning venues; some lead to full conference publications. The program hosts roughly seven interns per year. Research areas covered by mentors include reward learning, adversarial robustness of LLMs, interpretability, inverse reinforcement learning, cooperative multi-agent systems, probabilistic reasoning, generalization and misgeneralization, and the societal impacts of AI recommendation systems. The program is 12-16 weeks in duration, with flexible start dates throughout the year (though summer is most common). In-person participation at UC Berkeley is preferred, with remote options available for eligible applicants. Compensation is $5,000 per month for in-person interns and $3,500 per month for remote interns. International applicants are accepted. The application process involves four phases: initial application review, a programming assessment, two rounds of interviews with potential mentors, and an offer stage. Funding for the internship program is provided in part through the Berkeley Existential Risk Initiative (BERI) with support from Open Philanthropy. CHAI itself has received over $17 million in funding from Open Philanthropy as of 2022 and is also supported by the Future of Life Institute, The Leverhulme Trust, and CITRIS. Alumni of the program have gone on to PhD programs at top universities and positions at leading AI research labs including DeepMind.
Theory of Change
The CHAI Internship builds the AI safety research talent pipeline by training promising individuals in technical AI safety research methods early in their careers. By giving students and professionals hands-on, mentored research experience and publication credits, the program aims to increase the number of skilled researchers working on alignment and safety problems. CHAI's broader theory of change holds that provably beneficial AI requires fundamental advances in value alignment — particularly inverse reinforcement learning and related techniques — and that a larger, better-trained research community will accelerate progress on these problems before advanced AI systems pose existential risks.
Details
- Start Date
- -
- End Date
- -
- Expected Duration
- -
- Funding Raised to Date
- -
- Last Updated
- Apr 3, 2026, 1:21 AM UTC
- Created
- Apr 3, 2026, 1:21 AM UTC