Everything written about AI verification says check more. Almost nothing says where checking stops paying — and it does stop.
- Verification that costs more than the task means you should have done the task.
- A checkpoint that fires constantly trains you to approve without reading.
- Checking everything equally means checking the expensive things at the same depth as the cheap ones.
- The failure this produces is not a bad output. It is abandoning a tool that was working.
Checking AI output is necessary. Checking all of it, all the time, at the same depth, is not.
It costs more than the work saved, it exhausts the attention you need for the parts that matter, and it makes people rubber-stamp.
The arithmetic nobody does
A task takes you 30 minutes. The model does it in 2. Verifying properly takes 25.
You saved three minutes and took on the risk of a mistake you might miss. That is not a win, and it is extremely common in exactly the tasks people are proudest of automating.
If verification costs more than doing it, you have not saved time. You have relocated it and added a failure mode.
The honest test, before automating anything: how long does checking take, and is that number smaller than doing it? If you cannot answer, that is the finding.
Review theatre
The specific failure of over-auditing is a checkpoint that fires so often it stops being read.
An agent that asks permission for every step trains you, within about a day, to click approve without looking. The checkpoint is still in the process. It is no longer in your attention, which is the only place it was ever doing anything.
Which is why automations puts one checkpoint at the last reversible moment rather than five throughout: one checkpoint you read beats five you do not.
The cost of checking everything equally
Attention is finite. Spending it uniformly means the tier-4 formatting gets the same scrutiny as the tier-1 irreversible send.
Uniform checking is not thorough. It is unsorted — and unsorted attention systematically under-serves the parts that matter, because those parts are rarer.
The tiers are on auditing AI-generated work. The point of them is not to check more. It is to check less, in the right places.
The failure nobody counts
Over-auditing has an outcome that never gets attributed to it: people conclude the tool does not work and stop using it.
They were not wrong that it was not saving time. They were wrong about why. The verification design was the cost, not the model — and because the abandonment looks like a sensible judgement, nobody revisits it.
Where to actually spend the attention
- Everything irreversible or public. Fully, every time, no sampling.
- Every specific claim — names, dates, figures, citations. Structurally the highest failure rate.
- Anything executable — but by running it, not by reading it.
- Nothing else, at that depth. Sample the rest and accept that some formatting errors will ship.
And one habit that replaces a great deal of checking: require evidence rather than assertion. A diff, a list, a count. An artefact you can scan in five seconds does more than a review you will not do.
Under-checking ships errors you did not catch. Over-checking ships errors you stopped being able to catch, because the attention was already spent.
Both look like diligence from the inside. Only one of them is.