Do AI Benchmarks Measure Real Work? 21 Leaderboard Evaluations Mapped to 41 O*NET Work Activities (2026)

We mapped the 21 evaluations Artificial Analysis runs to the 41 O*NET work activities: the Intelligence Index tests 8 and puts 73% of its weight on three, and no evaluation tests communicating with coworkers, a core activity in 541 of 911 occupations