Job description
Peer-Vetted Research Initiative for Safety Methodologies • Remote, Global
Remote, Global
-
In this 16-week fellowship, you'll conduct peer-reviewed AI safety research in interpretability, governance, evaluation.
-
Collaborate with 3 fellow researchers to surface papers and develop methodologies during weeks 1-3.
-
Participate in anonymous peer voting to select the strongest research approach for your team.
-
Execute intensive research sprints with mentor oversight, iterating on results and refining outputs.
-
Prepare final research for publication at top conferences or as open-source software.
Peer-Vetted Research Initiative for Safety Methodologies PRISM) runs a fellowship producing peer-reviewed research in interpretability, AI governance, evaluation, and alignment.