What every medical affairs leader needs to know about generative AI in pharma

Every industry is buzzing about generative AI’s potential, but medical affairs in pharma face unique challenges that demand tailored solutions. As a leader in this field, you’re likely wondering: where do we begin? How can we make an immediate impact? And what practical steps can we take to integrate generative AI into our operations?

We recently hosted a webinar featuring a panel of experts. The insights they share will provide the answers you’re seeking and inspire you to implement AI within your organization:

• Rob Stevens, 20-year veteran in the pharma industry, including former Global Head, Digital Medical Affairs at Novartis
• Zayed Yasin, MD, healthcare industry lead at Writer and practicing ER physician

Challenges in implementing generative AI

The main challenge with generative AI in pharma is figuring out how to strike the right balance. Responsible AI use is essential to avoid biases and hallucinations (false outputs) — but it often moves too slowly to make a real impact. Moving too quickly doesn’t work for the healthcare industry, where guardrails, accuracy, and regulatory compliance are major needs.

Legal departments emphasize their risk aversion due to the high stakes nature of healthcare, and rightfully so. Others in the organization grapple with the dual fears of job elimination because of AI automation and the potential for rogue AI behavior. This apprehension slows AI adoption, as employees worry about job security and the uncontrollable aspects of AI.

The historical underperformance of digital initiatives in medical affairs and the broader pharmaceutical industry also amplifies these concerns. Digital strategies have yet to fully deliver on their promise. But Rob Stevens thinks generative AI can change that.

“If we step back in time a little bit with medical affairs’ digital transformation, it’s clear that medical affairs was very late to the digitalization space,” Stevens says. “And in my opinion, it hasn’t really matured in such a way where there’s a measurable ROI on those digital efforts. I think it’s important to recognize that, because generative AI actually presents an opportunity for medical affairs organizations to course correct the lack of value creation that the earlier wave of digital tactics presented.”

Four practical steps for immediate implementation

Progress doesn’t have a pause button, and your organization can’t afford to lag behind. If you’re eager to integrate generative AI but unsure where to start, our panelists have outlined four practical steps to start your journey.

1. Familiarize yourself with AI

Championing AI adoption within your organization becomes much easier when you’ve had firsthand experience with the tool. Stevens recalls a conversation with a colleague who was eager to implement a chatbot strategy but hadn’t personally engaged with the technology.

“I asked the person, ‘Have you ever used a chatbot?’ And they said no,” Stevens says. “If you haven’t used [generative AI], you should be really trying to learn it and understand it so you can better articulate the benefits. I find many people I talk to will talk about wanting to do generative AI in their organizations, but have actually never really dabbled in it.”

Hands-on experience with generative AI not only debunks fears and myths but also builds the trust, credibility, and confidence necessary for overcoming the barriers to organization-wide adoption.

2. Engage stakeholders early

Bring IT, legal, and compliance on board early in the ideation process for use cases. One of the worst things you can do, Stevens cautions, is develop a Medical-Legal-Regulatory (MLR) review use case for generative AI, and then go get approval for it.

“It’s really important that you ensure that the internal stakeholders are managed,” Stevens says. “Bring them in, have them be a part of it. It’s cultural, it’s change management 101.”

Not only will this allow you to build trust within the organization, but it can help make sure that you develop AI solutions responsibly.

3. Start small and build trust

In pharma, there’s no room to dive into the metaphorical deep end with generative AI. You need to wade in carefully. The moment generative AI doesn’t deliver or perform as it’s supposed to, it may start losing credibility, Stevens warns. Start with low-risk, operational use cases to demonstrate value.

“There’s some really exciting things that you can do with generative AI, but they’re risky and complex,” Zayed Yasin explains. “It takes you one, two, three years to be able to validate it before you feel safe putting it out into the world, and it’s going to be hard to hold onto the budget. It’s going to be hard to convince people this is really worth doing.”

Organizational digital strategies are bi-directional. They either exist to make an organization money, or to save an organization money. Keep this in mind when proposing use cases, as showing tangible value to your stakeholders is crucial to building trust.

“I’ve made mistakes in my career when I look at digital as a flashy new object. The reality is it has to work to accomplish one of those two aims,” Stevens says. “In medical affairs — especially as enterprises struggle with profitability — generative AI is one where I think you can actually prove out ROI.”

Find big, “boring” use cases that establish goodwill. Begin with small, operational projects that have the potential to scale. Consider, for example, contract processing — from CRO agreements to consulting contracts. This is a great starting point for generative AI, as it directly involves legal, reduces administrative burdens, and presents minimal risk. These areas can show quick ROI and build organizational confidence.

4. Measure and assess

Align your use cases with ways you can effectively measure impact. For contracting, calculate the current cost per consultancy agreement by considering total department resources and output volume. After implementing a generative AI solution, assess again. The end goal should be to lower the cost per output.

“I think where medical affairs have struggled with digital is being able to articulate what the benefit is,” Stevens says.

In pharma, swift responses are crucial, especially when patient health is at stake. If a doctor’s query takes a week to answer, engagement fades and they lose the moment. Delivering information when it’s most needed amplifies its impact — which is exactly where generative AI can make up for lost time.

Use cases where generative AI can have the greatest impact

The best use cases are organizational-dependent. Think about where you have administrative burden and identify use cases around that‌. By demonstrating value and aligning stakeholders, you’ll cultivate organizational confidence in generative AI.

Field medical (MSL) administrative tasks

Medical Science Liaisons (MSLs) are an expensive function, yet they spend roughly 8–10 hours a week on administrative or time-consuming tasks, like documenting interactions with key opinion leaders. From a business perspective, it makes sense to help the organization’s medical talent reduce time on administrative tasks by using generative AI to accomplish them faster or more easily.

“These are low-value-add tasks that an MSL should not be doing,” Stevens says. “They should be focusing on customer engagement or driving scientific narratives.”

MSLs can also use generative AI to prepare for presentations. For example, they can ask generative AI to review a presentation and present potential questions and objections from healthcare professionals.

Medical content/literature insights

Insights aggregation is another example that’s low risk, can prove ‌value, and solves a very real business problem.

“People really haven’t cracked the code on how to smartly manage insights, especially with aggregation integrating with RWD (real-world data),” Stevens says. 

Generative AI can help medical affairs manage and aggregate insights by providing tools to efficiently process and summarize large volumes of information. It can create plain-language summaries, extract key insights, and organize them in an accessible and actionable manner.

Medical, legal, regulatory review

Generative AI can also assist with MLR review.

“You can train your applications for the ABPI, the Swedish Medical Association, or the FDA and wind up having a particular compliance approach for every different jurisdiction,” Yasin explains. “No one human could keep all of that in their head or even know how to reference for it.”

MSLs can use a custom-built app from Writer to respond quickly and accurately to Medical Inquiry Letter (MIL) requests from prescribing physicians.

Future of generative AI in medical affairs

We’ve gone over what you can do in the present, but what does the power of generative AI hold in the next decade? For Stevens, he sees a future where AI operates in a more agentic model — meaning AI that can act autonomously.

“I see a future where an MSL has essentially a digital twin, where certain tasks that the MSL used to do are put on autopilot…This is not happening in one, three, or five years from now,” Stevens clarifies. “But I do see it where ultimately you won’t need as much headcount in a field medical function.”

This isn’t about replacing workers — it’s about redefining roles. Organizations will need to rethink resource allocation and strategically repurpose talent. Ultimately, human and AI collaboration will be key — it’s about people and machines working together, not against each other.

With this in mind, Yasin stresses that every role is going to get more strategic.

“​​This is the time for all of us to lean into some of this discomfort, but also think more innovatively about what the future roles look like.”

Expect future clarity with regulation and legislation simplifying AI adoption. But don’t wait — many pharma companies are already fully embracing generative AI. And how can you join their ranks?

“A pharmaceutical organization needs to have a unified generative AI platform,” Yasin advises. “Every organization and every function within an organization have five, 10, 20, 30 different things that they need done where generative AI can be of huge value to them. And if you have to go out and buy one point solution for each one of them, it quickly becomes ungovernable.”

For more expert insights, watch the entire webinar recording. Dive deeper into implementing generative AI at your organization and discover how Writer can partner with you on this journey.