A young man with short light brown hair, fair skin, and blue eyes, wearing a black shirt, looking directly at the camera against a plain beige background.

Lewis Hammond

  • Area: Cooperative AI

    My projects focus on multi-agent safety, cooperative AI, and/or governing AI agents. Several projects are also relevant to scalable oversight. They range from conceptual/theoretical to empirical, and include a couple of AI governance/ethics projects for those with less technical backgrounds. Please feel free to contact me for access to a non-public database of around 15 projects that I am currently interested in mentoring. I am also happy to consider proposals from mentees. The more specific themes of my proposals are:

    • Detecting collusion, the emergence of 'collective agents', and novel, dangerous collective goals or capabilities;

    • Analysing the importance of differing capability levels in strategic interactions between AI agents;

    • Understanding how to measure the 'cooperative intelligence' of AI agents, as well as differential progress on cooperation;

    • Creating new proposals for governing AI agents, potentially inspired by existing domains such as trading algorithms in financial markets.

    If you have research ideas in these areas, or more generally in multi-agent safety, cooperative AI, and/or governing AI agents, then please feel free to get in touch with me.

  • Lewis is the co-director of the Cooperative AI Foundation and a DPhil candidate in computer science at the University of Oxford. He is also affiliated with the Centre for the Governance of AI and a ‘Pathways to AI Policy’ Fellow at the Wilson Center. His research concerns safety and cooperation in multi-agent systems, motivated by the problem of ensuring that AI and other powerful technologies are developed and governed safely and democratically.

  • Each of my listed projects describes the background I would expect a mentee to have, and the difficulty level of the projects along different axes. In general, I would expect scholars to be interested in multi-agent problems and have some relevant background (e.g., having taken a course on game theory, or an equivalent level of self-study). Applicants should also be reasonably autonomous, proactive, and good at managing their own time and projects (though I understand that Pivotal provides some assistance in this regard).

Research Director, Cooperative AI Foundation