Demystifying AI Adoption for Medical Writing: An Interview with Núria Negrão, Part 2

This is the second part of our interview with Núria Negrão, a medical writer and AI adoption strategist specializing in continuing medical education. If you haven't read part one, we suggest you start there.

What would you say are some of the biggest barriers to AI adoption in medical writing, and CME in particular? 

The first one is simply a lack of knowledge. Generative AI is still very new, and most people don't yet feel comfortable these tools yet. There is often some fear and resistance to new technology. And a lot of these anxieties are completely valid–for example, AI hallucination really is a problem. The way to overcome these anxieties is to realize that even if you do use AI in your writing, it doesn’t mean that the AI is doing the whole job. There’s still a huge role for the human being to play. And ultimately, you are the one responsible for the work the AI produces. 

Once medical writers start to use AI, they often encounter a learning curve. It takes time to get to a level where you begin to see returns on your effort. I often hear people ask, "But does it save time?" Medical writers need to see a real benefit to using AI if they are going to put forward the effort to learn how to use it. Additionally, a lot of the AI tools on the market today that are marketed for medical writers are quite expensive. So I often get asked questions like, what’s actually worth investing in? How do I know which tool to pick? And how much money is reasonable and how much is too much? These aren't easy questions to answer.

Another big issue is the problem of security. In CME we typically sign NDAs before we start a project, and so users need to feel confident that they can use AI tools in ways that won't violate their agreements. Medical education companies are wary of exposing proprietary information, so there needs to be a clear understanding of what does and does not violate a contract. Furthermore, not all CME providers have defined clear AI use policies for contractors, which can make using AI feel extra risky.

And finally, there is the problem of copyright. I often get asked questions like, how do I make sure that I am not violating copyright? What can I upload into a large language model? And even if I am not violating copyright, how do I know that the tool itself is following intellectual property laws? 

I think that many of these concerns will be addressed over time as AI becomes more ubiquitous and regulatory bodies catch up, but right now there are some real question marks around best practices.

Given all the concerns you just raised, how do you help your clients navigate their fears around AI?

I try to address them a few different ways, depending on the source of the fear. For example, if someone approaches me with concerns that the AI will produce a subpar product, I usually offer a reminder that the user is still in control. If you don’t like what the AI produces, you don’t have to use it. And you can always correct and improve the output too. In general, I think education is the best way to overcome fear. Being given a framework for a process or workflow and understanding its limitations often goes a long way toward alleviating AI anxiety. 

Another thing that often helps is to engage in open conversations about the future of the field. I try to be very open about this topic because now that AI is here, there is no putting it back. The alternative is ignoring it and being caught blindsided. Instead, I try to help medical writers find new strategies to market themselves and identify exactly what they bring to the table that clients still very much need. 

Let's turn to a more positive topic--what excites you most about the adoption of generative AI for medical writing?

What excites me the most is the possibility of being able to do things that weren't possible before because of time constraints or because of resources. I think the most exciting thing is actually that I don’t know what those things will be yet, because we have never had an opportunity like this before. But I’m excited to see what’s possible. 

Consider the invention of the internet. Could we have imagined the changes it would bring before it existed? Of course there are downsides to the internet as well, but very few people would say they would rather it didn’t exist today. I think something similar will occur with AI. It has already allowed me to do so many things that I would not have been able to do before, like making images, videos, even developing small applications. Learning to do that would have taken too much of my time and resources in the past, but with AI suddenly these things are possible. When you scale that up, suddenly a lot of people are capable of so much more than they were in the past, and I think that’s really exciting. 

There certainly are some democratizing features of AI that are opening doors to new skills for many people. Let's hope that we are able to maximize those qualities while minimizing risks. That said, are there any specific models or tools you would recommend to newbies who want to start experimenting with AI?

Yes, definitely! For newcomers, I would recommend using Perplexity and Notebook LM for research and summarizing sources. If you feel comfortable, try uploading papers to Notebook LM and start asking it questions. Ask it to create a podcast or a video for you that summarizes the content you uploaded. Most people find that Notebook LM really speeds up their ability to become familiar with a topic.

Perplexity is good beginner tool because it can get answers with clickable citations to really hard questions that were impossible to ask an AI before. That’s a great way for a medical writer to start to see the benefit of using AI tools for research. Just play around and have fun, and then you can start to figure out what works for you!