humancompatible.ai: Advancing Human-Centered AI Research

humancompatible.ai is the official website of the Center for Human-Compatible AI (CHAI) at UC Berkeley, dedicated to reorienting AI research toward systems that are provably beneficial to humanity. With a mission rooted in technical AI safety and interdisciplinary collaboration across social sciences, CHAI pioneers foundational work in areas like assistance games, multi-agent cooperation, and ethical AI design. The platform serves as a hub for cutting-edge research, policy influence, and public education, featuring newsletters, progress reports, blogs, and academic publications. It supports global efforts in AI regulation, including contributions to the EU AI Act and U.S. congressional discussions, and fosters innovation through workshops and student research.

Key Features:

• Research Hub: Comprehensive access to CHAI’s latest research in human-compatible AI, including assistance games, adversarial robustness, and ethical design.
• Policy Influence: Active engagement with governments and international bodies (EU, US Congress, OECD, UNESCO) on AI regulation and safety standards.
• Quarterly Progress Reports: In-depth updates on CHAI’s research milestones and impact, published since 2017.
• Academic Blog: Insightful articles on AI ethics, political neutrality, reinforcement learning, and societal implications by leading researchers.
• Newsletter & Mailing List: Regular updates and thought leadership delivered quarterly to subscribers via MailChimp.
• Global Workshops: Hosts high-impact events like the NSF Convergence Accelerator Workshop on Provably Safe and Beneficial AI and the EDAI workshop on ethical AI design.
• Student Research Output: Publishes 32 student research papers annually, showcasing emerging talent and innovation.
• Public Education & Advocacy: Provides accessible content to inform the public and guide responsible AI development.
• Collaborative Network: Partners with leading institutions including the Leverhulme Centre for the Future of Intelligence, CITRIS, and the Berkeley Existential Risk Initiative.

Pricing: humancompatible.ai is a non-profit research initiative supported by donations and institutional funding. It offers free access to all research, publications, and public resources, with no paid tiers or usage fees.

Conclusion: humancompatible.ai is a vital, trusted resource for advancing safe, ethical, and beneficial AI, making it an essential platform for researchers, policymakers, and anyone committed to shaping a future where AI serves human values.