Judge feedback

Judge feedback is anonymous.

• I like this! Creating a public ranking for AI systems' safety or control would be great and might serve as a process for shifting incentives in for-profit AI development. There's great potential in e.g. doing this in collaboration with the European Union to control public procurement contracts to be above a certain threshold on this public ranking list to actually change the incentives monetarily as well. AI safety certificates are being pursued by several organizations, researchers and certification companies and I imagine that you could have a serious positive impact by trying to prototype a sort of ranked listing based on public benchmarks for safety. Go make it happen!