Who Made the Call
Short answer: AI advice is most useful inside the work, at the moment of decision, not in a report before or an audit after. A London surgical team showed the arrangement: the system marks the risk on screen, the surgeon decides. Organizations can copy it by naming, for each AI check, who acts on its flags.
What happened in the operating room?
At the National Hospital for Neurology and Neurosurgery in London, surgeons removed a pituitary tumor from a 48-year-old patient, Rhys Hibbert, while an AI system analyzed the live video from the operating camera and highlighted the critical structures at the base of the brain. It was the first operation of its kind, part of a clinical trial funded by the NIHR and Google. The system was developed at University College London and trained on hundreds of annotated videos from earlier operations.
The surgeons stayed in control throughout. Hibbert, who had been losing the lower part of his vision, woke up able to see clearly.
Why does the placement of the advice matter?
Most organizations deliver AI advice at a distance from the decision. An AI review of a contract arrives two days before the negotiation, or an error check runs after the invoice has gone out. Both have value, but neither is in front of the person at the moment they choose.
The surgical arrangement is different. The warning appears on the same screen the surgeon is already watching, while the instrument is moving. The person accountable for the outcome sees the risk at the point of action, and nobody has to recall a review from two days earlier.
The design also leaves no doubt about authority. The system marks; the surgeon decides. Sophia Bano, the technical lead, described the value as experience: the system has seen “a breadth of surgical examples that would take a surgeon many years to encounter.” It lends that experience to the person in charge. It does not replace them.
How can a team apply the same design?
Move checks into the work. Where AI reviews drafts, contracts or budgets, have it annotate the document itself, with the reason next to each mark, rather than producing a separate report.
Assign every mark an owner. Decide in advance who acts on the flags for each kind of work. A flag that belongs to no one is read by no one.
Record disagreements. When the owner overrides a flag, a one-line note says why. Over time those notes show where the tool helps and where it misleads.
Where AI sits in your team’s decisions is a design choice. Design it deliberately →
Keep reading: What a Formula 1 pit wall shows about judgment and speed · The metric patients noticed was where the doctor looked
This is the thinking behind a Trusted AI Culture.
Sources: UCL News, "First patient in live AI-assisted sight-saving brain surgery," August 26, 2026. Futurism, "Doctors Just Used AI to Perform Brain Surgery," August 29, 2026.
Written with AI in the loop: my idea, AI drafted and sharpened, my judgment on the way out. Every word is mine to stand behind. — Darren