JobBench: Aligning Agent Work With Human Will
For AI safety and labor economics, JobBench provides a benchmark that shifts evaluation from economic replacement to human-centric delegation, though it is incremental as it follows existing benchmark paradigms.
JobBench evaluates AI agents on 130 tasks across 35 occupations, prioritizing tasks experts want delegated. The best model (Claude Opus 4.7 under Claude Code) achieves only 45.9% accuracy, highlighting a gap in aligning agent work with human will rather than economic value.
Current benchmarks for occupational AI agents are scoped primarily by economic values, telling a replacement story. We introduce JobBench, which evaluates AI agents on the workflows that experts identify as high-priority for delegation, empowering humans based on their needs instead of replacing them with GDP value. JobBench covers 130 agentic tasks across 35 occupations. Each task is packaged as a workspace of heterogeneous reference files, requiring the agent to reason through the cluttered information streams of real professional work. Outputs are graded by a fact-anchored chain of rubrics, averaging 35.6 binary criteria per task. We evaluate 36 models; the strongest, Claude Opus~4.7 under Claude Code, reaches only 45.9 %. We hope JobBench shifts the community's target labour-market effect from replacement to enhancement: building agents that do what humans actually want delegated, not only what is most economically valuable.