John Goldie is a retired GP and medical educator
Imagine a man in his 30s presenting with palpitations. Before attending, he has used a symptom checker, reviewed his smartwatch data, and read discussions online. He arrives with a list of possibilities — anxious but also informed.
The signal is uncertain, but the meaning is already partly made.
It would be easy to respond by correcting — to reassure, to close things down. More often, though, we are working with interpretations that have taken shape before we enter the room. The task is no longer simply to replace his account with ours, but to understand how firmly it has formed — and how much effort it may take to shift it. He brings not only symptoms, but the interpretations that have gathered around them. The consultation becomes an effort to restore proportion: to decide which signals matter, which meanings can be stabilised, and which may still be open to revision.
The consultation becomes an effort to restore proportion: to decide which signals matter, which meanings can be stabilised, and which may still be open to revision.
This is now routine.
Patients often arrive having travelled through a dense informational landscape of devices, platforms, and forums. Their symptoms come already interpreted. More information does not necessarily clarify; it can also prematurely fix ideas in place.
General practice has always been less about certainty than interpretation. What has changed is when that interpretation begins. We now work with fragments: symptoms half-formed, stories already shaped, meanings arriving with prior authority. The consultation is not simply the application of knowledge, but the attempt to regain enough coherence for action.
Wearable outputs, online estimates, and AI-generated summaries may not constitute knowledge in any strict epistemological sense.1 But they function as knowledge in practice. They shape what patients attend to, how urgently they act, and what they take to be plausible explanations. A smartwatch alert can feel more immediate than clinical reassurance; an online forum more compelling than epidemiology. This is not because they are more accurate, but because they arrive earlier, with coherence and emotional force. This is knowledge that arrives early — early enough to shape what follows.
And this process often begins before the consultation itself.
A patient submits an online request about breathlessness, uploads a photo of their inhaler, and pastes text from an AI-generated summary. A receptionist reads it, asks a few questions, senses something not quite right, and brings the appointment forward. By the time the GP becomes involved, the story has already been shaped.
We call this administration, but it is also judgement.
Technology does not remove this complexity; it redistributes it. In many consultations, we are no longer the first to interpret illness.
Clinical reasoning is increasingly distributed across people and systems that are not formally clinical: shared inboxes, triage pathways, home monitoring, patient-generated data.2 These do not simply transmit information — they shape interpretation. What remains distinctive about general practice is not that we hold knowledge, but that we help it become usable in context.
This work is not only cognitive. It is relational. Patients do not arrive as neutral carriers of information. Their interpretations carry fear, hope, and prior encounters with digital explanations. To dismiss them risks missing something important. To accept them without reflection risks being led by accounts that may already be incomplete.
A simple question can shift the ground: What have you already found out about this? It acknowledges that sense-making has begun elsewhere. It also reveals something more unsettling: that the consultation is not where understanding is first formed, but where it first becomes visible.
It is tempting to ask whether algorithms threaten or improve practice. That framing is too narrow. The more significant shift is temporal. Clinical interpretation increasingly begins outside the consultation and often arrives with momentum. The GP’s task is less to compete with prior explanations than to recognise how strongly they have taken hold — and whether they can still be re-opened.
Illness has always been lived and interpreted, not simply diagnosed.3 Technology does not remove this complexity; it redistributes it. In many consultations, we are no longer the first to interpret illness. But we may be the first to recognise when those interpretations have become too stable to change — coherent enough to reassure, compelling enough to delay action, and resistant to revision.
Our task is not to compete with early explanations, but to gently re‑open them — enough to keep the story live, proportionate, and safe.
Deputy Editor’s note – see also: https://bjgplife.com/post-turing-clinical-relationships-how-ai-is-reshaping-patient-behaviour-before-the-consultation/ and https://bjgplife.com/yonderjul26/
References
- Topol E. Deep Medicine. Basic Books; 2019.
- Hutchins E. Cognition in the Wild. MIT Press; 1995.
- Kleinman A. The illness narratives: suffering, healing, and the human condition. New York: Basic Books; 1988.
Featured Photo by Mohammadreza alidoost on Unsplash
