Ideas
Nicole Junkermann on Who Is Accountable When AI Supports a Clinical Decision
A tool may prepare the ground, yet the judgement still belongs to a person. Nicole Junkermann on where responsibility sits.
Back to the blog
Some of the most useful questions about artificial intelligence are also the least dramatic. Nicole Junkermann keeps returning to one of them: when a piece of software helps a clinician reach a decision, who is responsible for that decision afterwards?
What the tool is there to do
The promise is easy to describe. A system could read through a long record, gather what seems relevant and lay it out in a form that a busy person can take in quickly. It may flag something worth a second look, or suggest what to consider next. Nicole Junkermann makes the point plainly in her short film: AI can prepare options, sort information and suggest next steps, but responsibility does not move to the machine. People still make the decision.
Why responsibility stays put
It is tempting to think that a very capable assistant must share the credit and the blame. Yet accountability is a human arrangement, built on the idea that someone can be asked why they acted as they did and can answer in their own words. A clinician can explain their reasoning to a patient, listen to an objection and change their mind. A score on a screen can do none of those things, however carefully it was produced. In her view the person making the decision remains accountable for it, and those who chose the tool and showed colleagues how to use it carry a share of their own.
What good practice could look like
None of this is an argument against the technology. It is an argument for arranging things so that responsibility stays easy to see. A few simple habits could help: a plain record of when a tool was consulted and what it offered; room for a clinician to disagree without a long justification; time set aside to ask whether the suggestions still fit the people being cared for. These are modest, administrative things, and they may matter more than any single clever feature. The Writing Desk has its own version of the same idea, that a short honest note is worth more than a long defensive one.
A British habit worth keeping
Britain has a long tradition of care built around named people, someone whose job it is to know, to decide and to answer. That tradition could be a real asset as these tools spread. It offers a ready-made way of asking the right question at the start, before a system is switched on, rather than at the end, when something needs explaining. It also fits the view of thoughtful AI in British innovation that Nicole Junkermann set out in August, in which the tools matter less than the judgement around them.
There is something reassuring in all of this. Better software may make a long shift a little easier, and it may give a clinician more time to look up from the screen and talk to the person in front of them. If so, the decision could stay where most people would want it to stay: with someone who can look them in the eye and explain it.
This article is a personal reflection and general commentary. It is not medical, financial, investment or professional advice.
More from Nicole Junkermann
For more from Nicole Junkermann, read why UK health tech rewards careful stewardship, Britain's quiet strength in satellites and automation and why smaller AI models suit British firms. Reinforcement learning and the value of patience makes a good companion piece.