Tag: ai_evaluations
-
Media Monitor: MIT Technology Review Reports AI Models Lag on Judgment-Heavy Tasks Despite Rapid Gains
MIT Technology Review reports METR's updated evaluation finds AI capabilities still improving exponentially on verifiable tasks, but models fall short of humans on open-ended, judgment-heavy work.
Science & TechnologyBusinessArtificial IntelligenceComputers and InternetEconomy MIT Technology ReviewMETRAI evaluationsrecursive self-improvementAnthropicOpenAIMITsciencetechnology