NLResearch Fellow, AI Interpretability
GoodfireUSA
Global catastrophic risksAI safetyResearchSoftware engineering
Why this is on the board
Large AI models are trained by trial and adjustment, without anyone properly understanding why they do what they do. As a fellow you spend three months on a predetermined research project meant to make this visible. The leverage is in your final report: if a method or finding works, other researchers can adopt it to understand models better and make them safer.
This board is in beta and this is our judgement. If you think we have it wrong, tell us. Give feedback · How we reason about it
About the role
A three-month AI interpretability research fellowship executing focused projects under mentorship from leading researchers. - Execute an assigned interpretability research project and produce a research report by programme completion. - Implement feedback from mentors while maintaining independent execution and fast iteration. - Requires strong technical background in computer science or machine learning with prior research experience. - Requires Python proficiency and ML frameworks; foundation…
Want to understand this first?
Start with our free course: about 100 minutes in six short units, at your own pace, with no sign-up. To go further, apply for the EA Fellowship: five weeks of reading and discussion in a small group. Also free, but selective.