What impact will AI have on biomedical research over the next few decades?
Answers to this question vary wildly. Some prominent scientists have expressed skepticism that AI will have anything more than a marginal effect on the future of research. In contrast, others have argued that we are approaching the “singularity,” wherein the pace of AI development will accelerate exponentially to create a superintelligence far surpassing human abilities, yielding cures to all human diseases in just a few years.
A middle path between these two extremes could be extrapolated from what has happened in the world of chess. In 1997, Deep Blue became the first AI to defeat the reigning world chess champion when it beat Garry Kasparov in a dramatic 6-game matchup in New York City. Since that time, AIs have been superior to humans in chess. However, for two decades after Deep Blue’s historic win, teams of humans and AIs playing together collaboratively (so-called “centaur” chess teams) were able to defeat the strongest AIs playing alone.
Humans and AIs bring different strengths to chess. Humans bring big-picture creativity and strategic intuition, whereas AIs bring extraordinary computational power. When those strengths are combined, they are highly synergistic. In the past few years, the top chess-playing AIs have gotten so powerful that it’s no longer clear they can be consistently beaten by centaur teams involving humans. However, for a solid two decades, the best chess in the world was played by centaur teams representing a collaboration between humans and AIs.
Biomedical research may now be entering its own “centaur era.” Over the next several decades, the highest levels of research may be performed by centaur teams of humans and AIs working together collaboratively, each bringing their unique strengths to help answer key biomedical research questions. Viewed in this light, AI in biomedical research is neither a nothing-burger nor a magical god, but rather a powerful set of new collaborators for human scientists.
Obviously, the chess analogy is imperfect. Biomedical research is much more complex than chess. Indeed, a single cell is more complicated than a thousand chess games. This is probably why AIs have been able to dominate chess since the 1990s, but only recently had a significant impact on biomedical research, an impact recognized by the 2024 Nobel Prize for the development of the AI known as AlphaFold.
If the centaur era truly lies ahead, then the most successful biomedical researchers of the coming decades may be those who learn how to collaborate with AI partners most effectively. Just as the best centaur chess teams were not simply grandmasters with computers, but players who knew how to ask the right questions of their machines, scientists will need to develop new skills in guiding, interpreting, and challenging AI-generated insights. The researchers who master this partnership most fully may ultimately define the next era of biomedical discovery.
Sincerely,
Randy Hall, PhD
President, ASPET
