Evidence item

Stanford reports superhuman benchmark performance alongside large failures on simple tasks, supporting a domain-specific rather than universal claim about AI intelligence. DIRECT_TEXT

Authenticity UNVERIFIED · extraction UNVERIFIED · independence group none (unreviewed) · created seq 10145

Source: The 2026 AI Index Report — Technical Performance · locator {} · excerpt hash sha256:293750ee0d1d76bfed79d8493804ef122834a7a85f96fb14c185e9f547c027db

AI capability is outpacing the benchmarks designed to measure it, and surpassing human-level performance. ... AI models can win a gold medal at the International Mathematical Olympiad but still can’t reliably tell time ... the top model read analog clocks correctly 50.6% of the time, compared with 90.1% for humans.