Augmented Intelligence in #MedComms

I spoke to a large audience of medical communications professionals earlier today on the inflection point we’re witnessing regarding the transformative role of artificial and augmented intelligence (AI) in the life sciences. Thanks to Peter Llewelyn for having me on!

?

I’m fortunate to have been involved in the authorship, writing and development of 3 key considerations in this space of late, including helping to architect the Medical Affairs Professional Society’s Generative AI White Paper, released on 9-28-23, the Healthcare Communications Associations’ AI Roadmap, released in early October 2023 and the International Society for Medical Publication Professionals AI Position Statement, which will likely be published in early November 2023.

?

These statements have one thing in common; they recognize that our collective work, and world, will be forever changed and improved by AI and that if you’ve been waiting for guidance to say this, that waiting time is over; it is now time to develop a policy or protocol encouraging responsible use, to seek out appropriate training and experimentation and to think about collaboratively working with AI to improve clinician and patient outcomes.

?

The discussion and questions on this presentation, with Peter Llewelyn, ranged a fair amount. I stand behind our discussion that encouraged stakeholders to experiment and experience GenAI directly, both professionally and personally. You can’t form a strong opinion of it and its capabilities until you do so not in your personal life and not professionally; you need to do so responsibly, but you do need to consider doing so and in the near term.

?

Some other questions that surfaced for which sufficient time didn’t exist in-meeting, explored things like where ChatGPT should be used for medical writing, the role of GenAI and data analytics in the future of MedComms, data security of AI tools, training courses in prompt engineering and generative in general, plagiarism, ChatGPT as a search engine, and who wins in the age of AI.

?

I’ll address some of those issues here for those who have additional interest. Using the consumer-facing ChatGPT for medical writing makes about as much sense as using a phonebook as a step stool. You could do it, but that’s not its intended purpose, and you only have yourself to blame when it collapses on itself and you end up getting hurt. If you want to leverage Large Language Models for medical writing purposes, such a thing can be done, but they have to be done, on purpose, trained for that intention, and with a specific domain in mind. Many companies are pursuing this at present. When you do this, many solely human knowledge pursuits, e.g. writing first drafts, summarizing existing content, surfacing recommendations and names of live sessions is faster with the model than with humans alone, but may require some additional human activity, for consideration, idea generation, validation and transformation.

?

We’re at an early point in the lifecycle of AI in medcomms, in life sciences, in society. At a not-too-distant future, AI will power everything, similar to how everything is powered by electrified now. Finding use cases for AI will be easy work in the near future, since everything will have an AI component either on the front-end or the back end in the days to come. ?

?

To get there, the biggest need is around upskilling, i.e. training, and to get training, going to a reputable, trusted source where content is readily available makes the most sense. There is a lot of good content at sources like MIT Sloan, Google, Stanford on the space at large or in prompt engineering specifically on YouTube. I prefer Conor Grennan’s content, as he and I are collaborating on a few initiatives, but everyone has their own preference.

ChatGPT, and all Large Language Models, is not search on steroids; better to think of it as a prediction engine... a platform that suggests what could or might be, whether that's the next word in your "autocorrect", a possible image in a graphic you are building or any other creative or knowledge output you might be engaging with. These content types or modalities are expanding greatly, and working, and partnering with this generative technology is a way of seeing what's possible, not just writing better copy.

If you focus on only one thing, it’s about digital transformation; our world at large, personally and professionally is starting to be transformed vis a vis AI. This doesn’t mean that we’re moving into a world where AI, robots and the like take over the world, but into one in which collaborative AI, copilot AI, AI as thought partner, becomes the norm and maybe 10%, 25% of what we do, everywhere is contributed by AI, partnering with us as capable, competent experts to achieve effects that are 10 times better, faster, stronger than either the human or the AI can achieve on their own.