← Risk register Form ORR-T · rev. 2026-08-11

Terms

Written to be read rather than to be unreadable. If something here is unclear, that's a fault worth telling us about.

What this is

A research project that scores 830 US occupations for exposure to current AI capability, against a published rubric. Scores are assigned by a language model following that rubric, with human review on selected pages. It is published so it can be checked and argued with.

What it is not

  • Not advice. Not career, financial, legal, investment or employment advice, and not a basis for hiring, firing, admissions or lending decisions about any individual. It describes occupations in aggregate; it says nothing about any particular person's job, competence or prospects.
  • Not a prediction with a date. We score capability exposure, not adoption. There is no cost-to-automate term, so nothing here tells you when, or whether anyone will pay to deploy it at all.
  • Not measured. The scores are structured judgment, not observation. We publish the reproducibility figures — including where two independent runs disagreed — so you can weigh them accordingly.
  • Not a judgment of worth. "COOKED" describes task exposure. It is not a statement about the value of the work or the people doing it.

Accuracy, and the absence of a warranty

The site is provided as-is, without warranty of any kind. We believe the figures are careful and we document their limits at length, but we do not guarantee they are correct, current or fit for any purpose, and we are not liable for decisions taken on the basis of them.

If something is wrong, say so in public — that's the correction mechanism, and it works.

Employers named in the dispatch

The dispatch summarises and links published reporting. An organisation appearing there means a publication reported something, not that we investigated or verified it, and not that we are making a claim about that organisation's conduct or intentions. Headlines and links belong to their publishers; the one-line summaries are ours.

Inclusion is not endorsement of the reporting. If an entry misrepresents what was reported, email hello@cookedindex.com.

Contributions

Field reports and votes are submitted voluntarily. By filing a field report you allow us to publish it, edited for length and clarity, without attribution. Don't include anything that identifies you or your employer if that would put you at risk — we can't unpublish something that has already been read.

We review every report before publication and may decline any of them. Votes are self-reported and unverified; treat them as sentiment.

Money

Some links to training providers are affiliate links, marked "partner link" with rel="sponsored". We may earn a commission if you buy through them. They are matched to an occupation's computed skill gaps, and no provider pays for placement.

Sponsor slots, if any are sold, are labelled on every card. Nobody can buy a verdict, a score, coverage in the dispatch, or omission from it.

Reuse

Quote the verdicts and the figures freely, with a link. Underlying data from BLS and O*NET is US public domain and yours to use directly. The scoring prompt, rubric and pipeline are in the public repository.

Wages, employment and occupation definitions are United States only, from BLS OEWS and O*NET. Reported deployments in the dispatch are worldwide, so a page may cite an employer outside the US alongside US pay data.