About MetaRoles

MetaRoles covers AI that does or assists the work of a human role: software and generative AI, AI agents, and embodied AI such as robots, humanoids and autonomous vehicles. The site is organised by role. Each occupation is keyed to its O*NET-SOC code, the standard US occupation classification, so a radiologist, a warehouse picker and a paralegal each have one place where their AI use cases add up.

Every case is graded

Every AI use case on MetaRoles carries one of four evidence grades. The grade comes from the sources, never from our own judgement, and a case moves up only when a new source supports it.

In production

A named organisation runs it in live work.

Pilot

A named organisation is trialling it in real work.

Prototype

Shown in a lab, a paper or a demo, not in real work.

Vendor claim

Stated only by the seller, with no confirmation from the organisation using it or from an independent source.

A vendor's announcement is never called a deployment until the organisation running it, or an independent source, confirms it. Studies funded by a vendor or a lab are labelled as such and never ranked alongside independent results.

Three formats

• Rankings sort one measurable thing, such as adoption, deployments, benchmark results, price or exposure, and are named for what they measure ("most-deployed", "fastest"), never "best".
• Features look at one question with sourced analysis: what changed, why, and what the numbers show.
• Explainers show how one thing works, from a coding agent's test suite to the SAE levels of driving automation.

How we source

• Primary sources first: regulators and government data, company filings, peer-reviewed papers, standards bodies, the makers' own documentation and reproducible benchmarks. Each page lists what it used, and every claim is footnoted.
• Vendor figures are labelled ("vendor-stated", "rated") and kept apart from independent measurements.
• Dated and versioned: every table says when its figures were checked, and statements about a product name its version.
• Trimmed, never padded: if reliable data runs out at 12 rows, the table stops at 12.

What we don't do

• Exposure isn't job loss. A score for how exposed a role is to AI, or how much AI is used in it, is labelled as exactly that. Job-loss forecasts are always attributed to whoever made them, with their method and date.
• We explain, we don't advise. Nothing on MetaRoles is medical, legal, financial, hiring or career advice for your own situation. Where we cover skills and tasks in a role, we report what the data shows.
• No named workers. Job cuts are reported as companies, filings and regulators state them. Public figures appear only in their official capacity.
• No news articles, reviews, opinion pieces or interviews, and no coverage of adult or sex-work AI.

Not affiliated

MetaRoles is an independent site. It isn't affiliated with any AI company, robot maker, employer, agency or researcher it covers. Names, logos and trademarks belong to their owners.