AI should augment Medical Affairs professionals — not replace scientific judgement
Every conversation about AI in pharma eventually arrives at the same anxious question: does this replace me? For Medical Science Liaisons, the honest answer is no — not the parts of the job that actually matter. But that answer only holds if we're deliberate about where AI sits in the workflow.
What AI is genuinely good at, right now
Generative AI is strong at compressing time spent on structured, repeatable tasks: triaging literature, drafting a first pass of a congress debrief, organising stakeholder research into a consistent template, or turning scattered field notes into a structured insight. None of that requires clinical judgement — it requires speed and consistency, which is exactly what these tools are built for.
What it isn't good at — and shouldn't be trusted with
AI doesn't know what it doesn't know. It can produce a confident, well-formatted, incorrect summary of a clinical trial. It can't read a physician's hesitation in a meeting and adjust the conversation accordingly. It can't take responsibility for a scientific claim. Those are exactly the moments where an MSL's judgement is the whole point of the role.
A simple test before adopting any AI use case
Before I bring an AI tool into a workflow, I ask one question: does this make my scientific judgement better, faster or more available — without ever standing in for it? If the tool drafts and I decide, it passes. If the tool decides, it doesn't belong in the workflow yet.
Where this leaves Responsible AI
None of this works without discipline: respecting privacy and confidentiality, verifying against primary sources to catch hallucinations, keeping a human in the loop, and staying inside your company's compliance and governance framework. I go deeper on each of these in AI in Medical Affairs.
The bigger picture
The Medical Science Liaison role isn't shrinking because of AI — the value of the parts only a human can do is becoming clearer. That's the shift worth paying attention to.
Explore further
See practical prompts and workflows built on this framework in the Medical Affairs AI Lab.