AI won't replace the MSL. But it is changing what makes the best MSLs valuable.
If you talk to medical science liaisons (MSLs) right now, the tension shows up quickly.
They are being told, explicitly, that AI is going to become part of their job. Not in a distant, theoretical way, but as something they are expected to start using now. At the same time, the core of the role is more important than ever. They are still expected to walk into conversations with clinicians and be right. Not approximately right. Not directionally right. Right in a way that scientific evidence holds up to questioning.
That combination is uncomfortable. Because the general AI tools that are widely available in most companies now don't operate to that standard.
What is changing now is that MSLs no longer have to choose between speed and scientific defensibility. A new generation of medically tuned, evidence-grounded AI tools is beginning to close that gap by bringing field insights, published science, and congress learnings into one traceable workflow. Instead of relying on generic summaries that smooth over uncertainty, MSLs can begin to work from outputs that show where the evidence came from, how it connects across sources, and what holds up under scrutiny. That changes AI’s role in the field from convenience tool to scientific force multiplier.
From Medline to ChatGPT
A few years ago, the workflow was slower, but clearer. You searched Medline, checked internal resources, pulled papers, read them, built your own view, and went into the conversation knowing exactly where each point came from. It took time, but it was defensible.
That is no longer how many MSLs are working.
Increasingly, they are doing what everyone else is doing: opening ChatGPT, Gemini, Copilot, or whatever sits behind their company firewall, and asking a question. Sometimes it’s simple: what’s happening in this disease area, what are clinicians struggling with, what themes are emerging from recent data? The answer comes back in seconds. It reads well. It sounds authoritative.
And that’s exactly the problem.
Unless you already know the space well enough to challenge it, you don’t know what’s been smoothed over, what’s missing, or what’s simply wrong. There’s a saying in AI research: AI is always confident, and sometimes right.
That risk plays out in very practical ways. An MSL walks into a discussion with a physician or KOL who knows the therapeutic area deeply, repeats a point that doesn’t quite hold up, and the dynamic shifts. You don’t get a second chance at credibility in those conversations. The reputation of the MSL role has always rested on a single premise: you can trust what an MSL says because it is grounded in evidence, and it is the latest evidence.
Two MSL camps on AI
Some MSLs are leaning into AI because they feel they have to. The message coming down from medical affairs leadership is clear enough: “AI won't replace you, but people who use AI will”.
Others are holding back, limiting AI to low-risk tasks like drafting emails or reformatting documents because they don’t trust it enough to use in scientific conversations. These tend to be career MSLs who have spent 15 or 20 years building clinician relationships they are unwilling to jeopardise.
Neither position is workable for long. What both groups are navigating is the same underlying problem: the tools most readily available to them were not built for what they actually do.
What the role was built on
Part of what has historically made the MSL valuable is exclusive access. They were the most efficient route for a time-poor clinician to get to the latest science, interpreted by someone who understands both the data and the clinical context.
That exclusivity is eroding. Clinicians are using LLMs themselves now – on their phones, at the end of a shift, in moments when they would never have reached out to an MSL anyway. But this is not the end of the MSL's value. It is a redefinition of where that value lives.
If the information itself is easier to access, then what differentiates an MSL is what they do with it: how they interpret it, connect it to what clinicians are seeing in practice, and confidently explain what holds up and what does not. That requires becoming the custodian of what quality and truth look like.
The advantage that's now available to everyone
Early in my career, I covered four countries as an MSL in an experimental European team where there was typically one MSL per country. That gave me something most MSLs didn’t have: cross-territory visibility.
I could walk into a conversation and share what was happening in France, what Germany was struggling with, and how the same data was landing differently across markets. That made me genuinely useful in a way an MSL covering a single geography couldn’t replicate.
That advantage shouldn’t depend on circumstance.
The same cross-territory insight that once required a network of colleagues and countless phone calls should now be available to any MSL with the right tools. Not just what they entered into their own CRM, but what colleagues across geographies are hearing, how it compares, and whether what a clinician just raised in Birmingham is a local outlier or something being reported across three countries. That changes the quality of the conversation entirely.
What congress data used to cost
The congress problem is a version of the same issue.
Anyone can read an abstract. The hard part is understanding how the data is landing: whether clinicians trust it, whether the endpoints feel meaningful, whether something about the design raises doubt.
Historically, most of that lived in conversations that weren’t captured, or were captured too late. By the time a congress report arrived two or three weeks later, assembled from scattered notes, the moment had passed.
But those real-time reactions are some of the most valuable signals in medical affairs. That’s where cutting-edge science is released for the first time, and where experienced clinicians say out loud whether an endpoint is credible, whether a dataset changes anything for them, or whether a trial design missed something important.
What is now possible
The shift that matters is not about replacing the MSL's judgement. It is about removing the friction around it.
When field insights, published evidence, and congress reactions can be brought together in one place, MSLs start to see things that weren’t visible before. They can ask whether something they are hearing locally is widespread. They can prepare for a conversation with a clearer sense of what matters to that specific clinician. They can connect what one clinician raises in a meeting to what HCPs across different markets have been saying.
That is a different level of insight, and it depends entirely on being able to trust the output.
This is where general-purpose LLMs still fall short for MSL work. Plausible-sounding answers can still be wrong, incomplete, or stripped of context that would matter to an experienced clinician. For MSLs, that gap is not acceptable. The source needs to be traceable. The evidence needs to be verifiable. The role isn’t becoming less scientific. It is becoming more so.
The MSLs who will be most effective are the ones who use these tools to extend their expertise without letting the tools replace it.
About the author
Vic Ho is a distinguished medical affairs professional with over 20 years combined experience in field and strategic medical leadership roles. Before becoming the global medical solutions lead for Sorcero, she held positions as worldwide field medical communications lead for cardiovascular at BMS and head of medical capabilities and excellence at Jazz Pharmaceuticals, as well as consulting for many companies’ medical affairs teams. Ho is known for her contributions to advancing medical strategy and field medical impact measurement and is an active voice in the medical affairs community, driving optimisation of insights management and fostering customer and patient focused approaches.
