Research talks
Specification, robustness and assurance problems in AI safety. IJCAI AISafety workshop, August 2019.
Discussion on the machine learning approach to AI safety (blog post). Effective Altruism Global London, October 2018.
Research problems in AI safety (slides, video). Oxford AI Society lecture, October 2018.
Impact measures and side effects (slides). Aligned AI Summer School, Prague, August 2018.
Measuring and avoiding side effects using relative reachability (slides). Goal Specifications for RL (GoalsRL) workshop, ICML, July 2018.
Interpretability for AI safety (slides, video). Interpretable ML symposium, NIPS, December 2017.
Reinforcement Learning with a Corrupted Reward Channel (slides, video). Workshop on Reliable AI, October 2017.
AI Safety: what, why and how? (slides). AI and Society conference, October 2017.
FLI talks
Highlights from Asilomar workshop on beneficial AI (slides, video). Beneficial AI conference, January 2017.
AI safety: past, present, and future (slides, video). Cambridge Conference on Catastrophic Risk, December 2016.
The story of FLI (slides). Effective Altruism Global X Oxford, November 2016.
AI risk: why and how? (slides) Governance of Emerging Technologies Conference, May 2016.
Introduction to global catastrophic risks (slides). Effective Altruism Global X Boston, April 2016.
