Yhn CultureTelevision first seen 2 d ago, last 28 min ago, peak #3

90% of US working hours poorly covered by AI benchmarks

Original: 90% of US working hours are spent on tasks not covered well by AI benchmarks

New research from Mercor maps AI benchmark coverage against US labor data and finds that roughly 90% of working hours in the United States involve tasks that existing AI benchmarks measure poorly or not at all. The finding is drawing attention because it suggests current evaluations may overstate how ready AI systems are for real-world work, and that benchmark suites lag behind the actual mix of human jobs.

Why now: People are debating whether AI progress and benchmark scores actually reflect the tasks most workers do.

MercorUnited StatesAI benchmarks

Open on hn →

Rank over time, top of the chart is #1. 175 snapshots from 2 d ago to 28 min ago.

Evidence

• 90% of US working hours are spent on tasks not covered well by AI benchmarks · speckx · 620

API: https://socialmediatrends-api.osmike.com/v1/trends/2510056