LO
Surgeon — tasks, one by one
The unit of analysis is the task, not the job title. Each one below carries its direction, whether the judgement rests on evidence or on platform inference, the reasoning, and what it does not establish.
Every task on this page#
Deciding whether to operate, and the consent
Still human-led✓ Evidence-backedWeighing the diagnosis, the patient's condition and what they want against the risks, and agreeing with the patient exactly what will and will not be done.
Risk calculators and imaging reports feed this decision, but the decision is a conversation with a person about their body, and the professional rules put it on a doctor: in the UK the doctor must be clear about the scope of what the patient is consenting to and must not exceed it except in an emergency.
The consent rule says who is responsible, not how much of the preparation — the information leaflet, the risk figures, the summary of the scans — software now drafts.
Planning the operation from the scans
Being augmented✓ Evidence-backedStudying CT and MRI, choosing the approach, and in some specialties building a plan that a robot or guide will follow.
In joint replacement the plan is increasingly built on a pre-operative CT and then carried out with a robotic arm, and England allows six such orthopaedic systems while evidence is gathered. The surgeon still decides the plan and approves what the robot will do.
The planning evidence is orthopaedic; in soft-tissue surgery the plan still lives mostly in the surgeon's reading of the scans, and nothing here measures planning time.
Operating by hand and through the laparoscope
Still human-led≈ Platform inferenceThe open and keyhole operation itself: dissecting, controlling bleeding, removing or repairing tissue, and closing.
The hand skill itself is not being automated; what is changing is how much keyhole surgery moves to a robot's console. Tissue that bleeds and moves, anatomy that differs from the scan and emergency operations are why the hands-on operation remains the surgeon's.
This rests on the nature of the work and on the robots being console-controlled; it does not say how quickly open and laparoscopic cases are being converted to robot-assisted ones.
Operating from the robot's console
Being augmented✓ Evidence-backedSitting at a console away from the patient and controlling the robot's instruments, with a team at the bedside.
Robot-assisted operations in England grew from about 3,099 a year in 2011/12 to 41,134 in 2023/24, and the systems in use let the surgeon control the instruments from a console — the robot does not operate by itself. In research, a robot has performed the clipping and cutting steps of a gallbladder removal autonomously in eight gallbladders outside the body, and its authors say a surgeon should always oversee it.
The growth figures are English and count procedures, not surgeons; the autonomous result was outside the body, without bleeding or breathing, and the surgeon in that study was faster.
What the scans did not show
Still human-led✓ Evidence-backedThe adhesion, the second tumour, the bleed, the anatomy that is not where it should be — deciding mid-operation whether to go on, change the plan or stop.
These moments are defined by falling outside the plan, and the rules make them a person's decision: consent guidance names the case where an opportunity for another intervention arises once an operation is underway, and bars going beyond what the patient agreed except in an emergency. At a robot's console the surgeon is away from the patient, which the NHS's own implementation report names as a challenge for situational awareness.
Nothing here counts how often plans change mid-operation, and the rule assigns the decision; it does not show whether decision-support tools are used in those moments.
Operation notes, letters and follow-up
Being augmented≈ Platform inferenceWriting the operation note, the discharge summary and clinic letters, and reviewing the patient afterwards.
Ambient scribe tools that draft clinic notes from the conversation are spreading in outpatient medicine, and templated operation notes are standard. The surgeon still reads, corrects and signs what is recorded, because the record is what the next clinician and any later review rely on.
No record here measures AI drafting of operation notes or surgeons' documentation time; this is an inference from how documentation tools are spreading in medicine.