If you are a doctor, nurse, or allied health professional looking for flexible remote work, health AI training and evaluation is one of the more interesting options available right now — and one of the least well understood.
This page explains what it actually involves, who it suits, and what to expect practically.
What the work involves
Health AI platforms like Mercor and Micro1 need clinicians to help train and evaluate their systems. That means reading clinical scenarios, assessing how an AI has responded to them, writing example answers that demonstrate good clinical reasoning, and flagging responses that are unsafe, overconfident, or misleading.
It is done online, task by task, at times that suit you. There are no shifts, no patients, and no coding.
The skill being used is the same one you use in practice — clinical judgement. The difference is that you are applying it to AI outputs rather than to patient presentations.
Is this a job?
No — and it is worth being clear about this upfront.
This is contract project work, accessed through platforms like Mercor and Micro1. You are not employed by this site, and we do not place you into any role. If you apply via links on this site, you will be contracting directly with those platforms.
Most clinicians doing this work treat it as supplementary or portfolio income alongside existing clinical roles — not a replacement for them. Work volume is project-dependent and can be intermittent, particularly early on.
Some platforms ask that you have a business bank account or limited company to receive payment. It is worth checking this before you apply.
Who it suits
This work suits clinicians who are comfortable reasoning through uncertainty, are cautious by instinct, and can articulate why something is clinically problematic rather than just that it is. It tends to suit generalists — GPs, acute medicine, emergency — particularly well, because the scenarios often mirror first-contact or undifferentiated presentations where safety-netting and appropriate deferral matter more than definitive answers.
Specialists do well when tasks are clearly scoped to their domain. The trap for specialists is defaulting to depth when the question actually calls for knowing when not to answer.
It is also worth saying explicitly: SAS doctors, IMGs, and locally employed doctors are often very well suited to this work. Platforms care about judgement and safety instincts, not training grade or job title.
You do not need coding skills, AI knowledge, or technical experience of any kind.
Fitting it alongside NHS or locum work
Most clinicians doing this work do it alongside existing roles. There is no fixed time commitment and the work is asynchronous, so it does not conflict with clinical schedules in the way a second job might.
That said, you should check your employment contract before starting. Most NHS contracts require you to declare secondary work, and some have restrictions. The key question is whether there is a conflict of interest — which is unlikely given this work involves no patient care and no identifiable patient data.
What to do next
If this sounds like it might suit you, the most useful next step is the self-assessment page — it will give you a realistic picture of whether your working style fits before you spend time on an application.
Written by
Sean Key
Digital Health Senior Programme Manager · 29 years’ NHS & private sector experience
Sean has spent nearly three decades delivering complex digital programmes across the NHS and private healthcare — from LIMS and PACS deployments to primary care, urgent care, mental health, and national interoperability work. Not a clinician. His perspective is that of a practitioner who understands how digital health really gets built, procured, and adopted in the real world.