NLResearch Scientist, Manipulation Evals
Apart ResearchRemote
Global catastrophic risksAI safetyResearchSoftware engineering
Why this is on the board
Regulators can only enforce rules for what they can measure, and whether an AI model manipulates people can still barely be measured reliably. A test you design here can be rerun by regulators and other researchers on every new model. That way, a single piece of research can determine what every subsequent model is tested against.
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
Lead end-to-end research on evaluations of harmful AI manipulation for regulatory and public use.
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.