- Who it helps
- Teams choosing AI evaluations and readers interpreting capability claims.
- Availability
- The map and methodology are publicly readable. The page’s exact publication day was not established. The accompanying original worksheet can be completed with a text editor and no paid account.
Coverage and capability answer different questions
A task appearing in a benchmark tells you that some related behavior is tested. It does not show that a model succeeds at the task, completes the surrounding workflow or saves a measured amount of working time.
Read the mapping assumptions
The methodology uses automated matching and a bounded expert check. Benchmark selection and the threshold for coverage affect the result; the expert check does not validate every proposed mapping.
What you can do
- Use the accompanying task-coverage worksheet to list a real workflow’s inputs, outputs and failure conditions before choosing a benchmark. Open reference ↗
Where the evidence stops
This is a coverage analysis with selection and mapping uncertainty. It cannot by itself estimate job automation, reliability or business returns.
Sources read. Sources checked October 17, 2026 UTC. No external product workflow, model inference or third-party project was executed.
Practical guide
Check whether an AI benchmark covers your actual work →
Finish with a task-coverage matrix, a list of untested behavior and a bounded decision about what evidence to collect next.
Sources & further reading
Web · paper
A benchmark map of the US economy ↗
Read the map’s coverage definition and limitations. The original page does not supply a reliable publication day; coverage is not a measure of model success.
Mercor · page checked 2026-10-17
Web · paper
Benchmark-map methodology ↗
Read benchmark selection, task matching and the limited expert check. The published mapping was not independently reproduced.
Mercor · page checked 2026-10-17
More sources reviewed (1)
Hacker News · discussion
Hacker News: ranked discussions ↗
First thirty ranked rows inspected, including benchmark coverage, research, models and code-review practices.
Hacker News · page checked 2026-10-17
Coverage note: Coverage is limited to the cited sources. Original X access, independent replication and a comprehensive platform comparison were unavailable.
The record over time (1)
Dated source checks and published explanations. An observation date can differ from the original release date.
Mercor maps a selection of AI benchmark prompts to occupational tasks. It helps readers ask which parts of a job an evaluation represents before treating a model score as evidence about the whole job.