NLCrucible Program
AE StudioRemote
Global catastrophic risksAI safetyResearchSoftware engineering
Why this is on the board
You put in about two hours a week, while an engineering team spends more than sixty hours and compute per sprint testing your idea. That way a hunch about how AI systems could become unsafe becomes, within two weeks, a tested result with a written-up analysis that other researchers can build on, even if the idea turns out to be wrong.
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
Propose and test an AI alignment research hypothesis with engineering support over two-week sprints. - Guide research direction with ~2 hours weekly input; receive $1,000 stipend per sprint. - Work with engineering team (60+ hours per sprint) to test hypothesis against one of six published research questions. - Access cloud compute, frontier model APIs, and receive documented results with structured analysis and IARPA recommendation. - Hypothesis must be testable, show initial signal within t…
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.