AI in the newsroom: the version that searches has stuck, the one that writes has not
Newsrooms are using it for checking rather than producing. The result is quieter than anyone predicted.
Two years ago the argument was about models that write copy. What newsrooms actually use now is duller: tools that go through archives, transcripts and datasets in seconds.
An editor working out of the studio puts it plainly — we already know how to write, it was finding the thing we were looking for that took the time.
On security the rule has not moved: the weak link is never the software, it is the user in a hurry.
Where it works
What decides how fast this spreads is not the technology but how long older handsets keep getting updates.
At the user end what changes is usually just the interface. The real change happens behind it, where nobody is looking.
Where it does not
Manufacturers tend not to comment on this sort of thing. By now the silence counts as information too.
Most of the experiments in having a model draft the piece have been rolled back. What is left is narrow but solid: archive search, translation and turning audio into text. Three invisible jobs, and all three save hours.
PremiumThe British-built model passes on short answers and comes apart on long documents
Solid on brief queries; on documents running to many pages the error rate climbs.
Leak: the new handset’s camera housing appears for the first time
The images are unconfirmed, but two separate sources show the same layout.
AI-assisted scan reading begins at three hospitals
The system does a first pass in radiology; the report is still signed by a doctor.