University of California, Berkeley Opens Applications for CHAI Research Fellowship 2026-2027. The Center for Human-Compatible Artificial Intelligence (CHAI) at the University of California, Berkeley is currently accepting applications for its CHAI Research Fellowship. This fully funded postdoctoral position supports innovative research in AI alignment, machine learning, decision-making systems, robotics, and the broader field of human-compatible artificial intelligence.
Hosted at one of the world’s leading universities for artificial intelligence research, the fellowship gives researchers the chance to work alongside internationally recognised experts while contributing to efforts that aim to keep advanced AI systems beneficial to humanity.
About the Fellowship
The CHAI Research Fellowship is designed to develop outstanding postdoctoral researchers who are dedicated to building safe, reliable, and human-compatible AI systems. Fellows work closely with leading figures in the field, including:
- Stuart Russell — a pioneer in artificial intelligence and one of the principal architects of the human-compatible AI research agenda.
- Pieter Abbeel — a leading expert in robot learning, reinforcement learning, and autonomous systems.
- Anca Dragan — a specialist in human-robot interaction and algorithms that enable robots to better understand and accommodate human preferences.
Fellows also have opportunities to collaborate with researchers from Berkeley AI Research (BAIR), other UC Berkeley faculty, and partner institutions around the world. The programme supports ambitious, independent projects that address some of the most pressing challenges in modern AI and machine learning.
Priority Research Areas
Given CHAI’s broad mission to advance beneficial AI, fellows are encouraged to pursue original and interdisciplinary work. Current areas of particular interest include:
- Theoretical foundations of intelligent agent architectures — exploring the fundamental principles that underpin how intelligent systems reason and act.
- Human-compatible AI systems — designing AI that can reliably interpret and act in accordance with human values and intentions.
- Decision-making over long time horizons — developing methods that allow AI systems to plan effectively across extended periods.
- Cumulative lifelong learning — enabling systems to continuously acquire and refine knowledge without forgetting previous skills.
- Human-in-the-loop planning — creating frameworks in which humans and AI systems collaborate effectively on complex planning tasks.
- Reinforcement learning — advancing algorithms that allow agents to learn optimal behaviour through interaction with their environment.
- Human-robot interaction — improving how robots understand, communicate with, and assist people.
- Probabilistic programming — using probabilistic models to represent uncertainty and support more robust reasoning.
- Formal reasoning systems — applying rigorous mathematical methods to verify the behaviour of AI systems.
- Human preference learning — developing techniques that allow AI to accurately learn what people want from limited or noisy feedback.
- Multi-agent cooperation — studying how multiple AI systems (or AI and humans) can work together effectively.
- Mechanism design for human-machine systems — designing incentive structures that encourage beneficial interactions between people and machines.
- AI alignment and control — ensuring that advanced AI systems remain under meaningful human oversight and pursue intended goals.
- Game-theoretic approaches to AI — applying insights from game theory to multi-agent and strategic AI settings.
- Control theory and intelligent systems — integrating classical control methods with modern machine learning approaches.
Researchers may also contribute to collaborative work that draws on mathematics, economics, computer science, robotics, statistics, and cognitive science.
Responsibilities
Fellows enjoy substantial freedom to shape their own research agendas while contributing to CHAI’s overall mission. Typical responsibilities include:
- Conducting original research in artificial intelligence and related fields.
- Publishing results in leading academic journals and conferences.
- Collaborating with faculty, postdocs, and graduate students.
- Leading or contributing to interdisciplinary research projects.
- Mentoring junior researchers and students.
- Supporting broader AI safety and alignment initiatives.
- Participating in seminars, workshops, and research events.
- Contributing to educational activities related to AI research.
The fellowship places equal emphasis on high-quality scholarship and leadership within the growing community focused on human-compatible AI.
Eligibility
Applicants should hold (or be close to completing) a PhD in a relevant technical field and demonstrate strong research achievements through publications. Suitable backgrounds include:
- Computer Science
- Statistics
- Mathematics
- Theoretical Economics
- Machine Learning
- Artificial Intelligence
- Robotics
- Data Science
Prior experience specifically in AI safety or alignment is not required, but candidates must show solid technical foundations and clear research potential. Familiarity with modern AI and statistical learning methods is considered an advantage.
Benefits
The fellowship provides comprehensive support so that researchers can focus fully on high-impact work. Benefits include:
- Fully funded postdoctoral support.
- An initial two-year appointment, with the possibility of renewal for up to three years based on performance.
- A competitive salary aligned with UC Berkeley postdoctoral scales.
- Close collaboration with world-leading AI researchers.
- Access to Berkeley’s advanced AI research infrastructure and community.
- Opportunities to publish in top-tier venues.
- Membership in an internationally recognised research network.
Visa Support for International Researchers
CHAI actively supports international applicants by providing:
- Visa sponsorship for successful fellows.
- Assistance with visa-related expenses.
- Contributions toward visa costs for dependents, including partners and children.
This support makes the fellowship accessible to talented researchers from around the world who wish to conduct advanced AI research in the United States.
Why Human-Compatible AI Matters
As artificial intelligence systems grow more capable and become more deeply embedded in society, ensuring that they remain aligned with human values has become one of the central scientific and societal challenges of our time. CHAI’s work focuses on developing AI systems that are demonstrably beneficial, safe, and responsive to human preferences and societal goals.
Research carried out under the fellowship contributes directly to progress in AI safety, AI alignment, responsible governance of AI, human-centred machine learning, ethical autonomous systems, and cooperative AI.
How to Apply
Applicants should submit:
- A current résumé or curriculum vitae.
- A one-page statement of interest describing proposed research directions.
- The names and email addresses of two academic referees.
Applications are evaluated on the basis of academic achievement, research excellence, leadership potential, technical expertise, and alignment with CHAI’s mission.
Fellowship Summary
- Programme: CHAI Research Fellowship
- Host: Center for Human-Compatible Artificial Intelligence (CHAI), University of California, Berkeley
- Location: Berkeley, California, USA
- Start date: Flexible
- Duration: Two years (renewable up to three years)
- Funding: Fully funded
- Visa sponsorship: Available
- Open to: International researchers
The CHAI Research Fellowship offers early-career AI researchers an outstanding platform to address some of the most consequential questions in contemporary science. With world-class mentorship, generous support, visa assistance, and a collaborative environment at UC Berkeley, it provides an ideal setting for those seeking to help shape the future of safe and beneficial artificial intelligence. VISIT HERE TO APPLY