LOWho makes this, and who pays for it
Every site that tells you what AI will do to your job is paid by someone. Knowing who is the fastest way to judge what you are reading — so that is the first thing on this page rather than the last.
What this is#
A task-level record of what has actually happened to specific occupations, with the sources attached.
Today that is 36 occupations broken into 186 separate task judgements, 12 university majors, and 90 verified events — each with a date, a link you can open, and a written note on what it does not establish. 81 of the 186 task judgements rest on one of those records; the rest are labelled platform inference on the page itself, per task, so you can see which is which without asking us.
How a judgement is made is written out in full at Method and evidence, including when we say we do not know.
Who makes it#
VOLO is run from Singapore by Wei Chuanjie. It is not a research institute and not a consultancy: it is one person, working with AI tooling that reads sources, drafts records and checks the site's own rules on every build. That is a small operation, and saying so is more useful to you than implying otherwise — you should read what follows knowing the size of the thing that produced it.
What that means in practice is that the discipline is written into the code rather than into a style guide. A record cannot reach a page without a reachable source and a named verifier. A task judgement cannot claim evidence unless a record that can move a baseline is attached to it. A page cannot ship if its next steps contain a platitude, or if its 'how long do I have' answer contains a year. Those are tests that fail the build, not intentions.
Who pays for it#
Right now: nobody. That is the honest answer, and it is also the constraint we intend to keep.
- No advertising, and no sponsored records.
- No affiliate links. Nothing here earns a commission if you click it, enrol in something, or buy a tool.
- No vendor pays to appear, and no vendor pays to be left out. Several records here are about vendors' products not working.
- Where a source has a stake in its own conclusion, the record says so in its scope — a vendor's own benchmark, a firm arguing for the thing it sells.
It cannot stay at nobody forever. What we intend to charge for is work done for one organisation — taking your actual roles and mapping them, task by task, against this evidence. What that is, and what it refuses to do. The content on this site is not what gets sold, and putting a task judgement behind a payment would break the only thing that makes it worth reading.
What we refuse to do#
These are not positioning statements. Each one is enforced by a test that fails the build:
- We do not answer 'how long do I have' with a number of years. Nobody can, and the confident ones are guessing.
- We do not score a major, and we do not rank occupations by replacement risk. A number invites you to stop reading at the number.
- We do not add capability, deployment and employment change together. A machine being able to do something, an employer putting it into production, and jobs changing are three separate questions.
- We do not tell anyone who to make redundant. Not on the site, and not in the paid report either.
When we are wrong#
We are sometimes wrong, and the useful thing is that it is checkable: every record names its source and its date, so an error here is an error you can point at. If a record misreads its source, if a scope claims more than the document says, or if something is simply out of date — write, quote the record id, and it gets corrected with the change kept in the record's own history.
hermans@foxmail.com — corrections, questions, or anything this site says that does not match what you see in your own work.