Raymond Douglas & David Duvenaud
This stream will explore AI slowdown dynamics, desirable futures, moral convergence, and open-ended forecasting.
Stream overview
Raymond Douglas
I'm pretty opportunistic; things I am currently thinking about which are representative:
- Should there be a field studying AI slowdown dynamics in general? What should it look like?
- What do good futures look like, and what are the toughest tensions? E.g. how do you handle paternalism? (I am personally sceptical of handover)
- How does moral convergence work? Is it a viable win condition?
———
David Duvenaud (co-mentor)
- Fleshing out what "political" natural selection will result in under various UBI schemes. I.e. how will people and machines Goodhart various attempts to redistribute wealth. Baby factories? Gerrymandering distributed compute to meet the minimal official threshold for consciousness?
- A more conceptual project outlining the factors pushing for agency at the global level, versus for uncoordinated competition. I.e. is the usual outcome for a civilization increasing centralization, or a runaway race to the bottom?
- Working on the technical problem of open-ended forecasting with LLMs, e.g. predicting future newspaper headlines. A relevant sub-problem is evaluating these abilities on historical data without being confounded by things like shifts in style.
Mentors
Raymond Douglas is a senior researcher at ACS Research, studying the societal effects of powerful AI systems and how agency works in AI. During the COVID-19 pandemic, Douglas published epidemiology research and advised the UK government.
David Duvenaud is an associate professor at the University of Toronto, researching AGI governance, evaluation, and catastrophic-risk mitigation. He previously led Anthropic’s Alignment Evaluations team.
Mentorship style
Fellows we are looking for
Project selection
The Winter 2026 cohort offers a wide range of research streams led by experts across AI alignment, interpretability, governance, and safety. Each stream provides its own research agenda, methodology, and mentorship focus.
