Very excited about all the interesting projects the Winter Fellows got up to.
Projects ranged from external scrutiny of frontier models, Chinese AI development, market concentration implications of foundation models, public input into AI, and more
Conversation
Joe O’Brien worked with me on a whitepaper that explores the role of external scrutiny in governing powerful AI models. Hoping to make it public soon.
2
6
Leonie Koessler explored risk assessment techniques that AGI companies can use to identify, explore, and compare existential risks from AI.
1
4
researched how information sharing between AI labs and the UK government can allow the latter to prepare for advancements in AI capabilities coming out of the former.
1
2
Conor Downey explored the US government’s legal authority to stop or delay frontier AI development, including the actions available to the government, the legal viability of such actions, and the costs and benefits involved with taking them.
1
2
produced an overview of China’s generative AI landscape, including key models, their capabilities, how Chinese actors engage with safety concepts, as well as opportunities and challenges faced by state actors and commercial developers in advancing China’s AI industry
1
2
Owen Yeung’s project highlighted the possibility (and desirability) of cooperation between the US and China on AI.
1
2
explored the question of where public input can play a role in AI governance, and how such input can be incorporated into decision making at AI labs.
1
2
Ben Bucknall investigated the question of “what access to models do AI researchers consider most important for different subfields of safety-relevant research?” by conducting an analysis of recent literature and semi-structured interviews with researchers.
1
2
analysed how the People’s Republic of China’s desire for increased geopolitical legitimacy plays a role in how they approach AI development and governance.
1
1
7
’s research project sought to answer what AI labs should do when they see worrisome safety results in competing labs, to which he proposes the idea of evaluation-based pausing schemes.
looked at the market structure for foundation models and what this implies about competition policy and regulation.
1
1
7
Looking forward to reading all these drafts they've produced and seeing all the exciting work they'll do in the future.
1
2
5
