On May 27, 2026, a new framework known as JobBench was introduced, which seeks to evaluate AI agents based on human workflows rather than traditional economic values.
Current benchmarks for AI agents often prioritize economic metrics, which may not accurately reflect the intentions and needs of human users.
JobBench aims to create a more aligned approach, focusing on how AI can effectively support and enhance human work rather than simply replacing it.