PRISM AI Safety Research Fellowship

• Conduct peer-reviewed AI safety research in interpretability, governance, and evaluation.
• Collaborate with a team of researchers to surface papers and develop methodologies.
• Execute intensive research sprints with mentor oversight and iterate on results.
• Prepare final research for publication at top conferences or as open-source software.

This fellowship advances AI safety by producing rigorous, peer-reviewed research that helps ensure AI systems are safe and beneficial.

• Strong background in AI safety, machine learning, or related fields.
• Ability to conduct independent research and collaborate in a team.
• Experience with research methodologies and academic writing.
• Familiarity with interpretability, governance, or evaluation of AI systems.

16-week structured fellowship with mentor oversight, collaboration with peers, and opportunity to publish at top conferences. Compensation details not specified.

The Peer-Vetted Research Initiative for Safety Methodologies (PRISM) is an organization dedicated to advancing AI safety through collaborative, peer-reviewed research.