This is an excerpt from a longer report on AI, medicines development and the patent system.
Download the full report here (or by clicking on the image at left).
Rapid recent advances in Artificial Intelligence (AI) are already having a transformative impact in the life sciences. Google DeepMind, a London-based AI firm, provides a well-known example. One of its AI systems, AlphaFold, came out of relative obscurity in 2018 to solve a fundamental problem in biology and medicine (the protein folding problem: predicting how sequences of amino acids fold themselves into biologically active three-dimensional protein shapes), earning its creators the 2024 Nobel Prize in Chemistry. AlphaFold and similar AI systems have the potential to have a transformative impact beyond fundamental science, however, in the much more commercial field of drug development and discovery. As one ex-DeepMind scientist recently said:
“In a perfect world, the dream of purely computational drug discovery comes to life,” he described. “You tell me, this is the sort of disease you’re going after. Maybe there’s a target protein that you have in mind. And at a push of a button, we can generate candidates that meet all of the criteria you care about, which de-risks the steps that come after, shaves time off the process, and at the end of the day, will yield better drugs, faster.” (“Ex-DeepMind scientist launches AI drug discovery venture”, Financial Times, 13 February 2025)
Such a perfect world could indeed be wonderful in terms of helping to solve many of the health-related problems that continue to afflict humanity (We charitably assume that this technology will be used for good rather than for ill.) However, thinking about the use of AI in drug discovery and development raises some nearer term thoughts about how such a change might impact the patent system which is, after all, the foundation on which the pharmaceutical industry business model rests.
Let us imagine that the use of AI in drug discovery does, as the above quote suggests, reduce the cost, risk and time taken for drug discovery and development. This would be hugely significant since the pharmaceutical industry has argued for many decades that countries must provide increasing levels of patent and other intellectual property (IP) protection, such as data exclusivity, to keep pace with the increasing costs, risks and time taken to discover and develop new drugs. Their success in persuading policy makers of this argument has been reflected in the ever-increasing levels of pharmaceutical IP protection around the world, part of a process which has memorably been described by Peter Drahos in terms of an ‘IP-ratchet’. If the use of AI in drug discovery were to reduce the cost, risk and time taken for drug discovery and development then, by the same argument, surely this should mean that levels of pharmaceutical IP protection ought to begin to be scaled back. Can they be scaled back, though, and, if so, how: might the IP-ratchet have a reverse gear after all? What will happen if they aren’t: could the pharmaceutical industry then look forward to the prospect of turbo-charging its profitability?
These are all important questions but, in fact, the situation might be even more complicated. As things stand, patent systems around the world do not regard an AI system as a valid inventor for the purposes of being granted a patent. It would therefore be ironic if the dream of purely computational drug discovery were to come to life but, as a direct result, the new drugs it discovers are unpatentable since no human inventor is involved. Alternatively, even if human inventors do remain involved, it could be that the widespread use of AI systems means that inventions such as new drugs become increasingly obvious over time, leading to those new drugs becoming unpatentable for lack of inventive step.
In either case, a lack of patents would be a huge challenge for the pharmaceutical industry business model. In the limit, it might be that vital areas of research and development become starved of pharmaceutical industry resources where the use of AI makes new drugs unpatentable. It seems unlikely that the pharmaceutical industry would passively accept this change, though, and it might be that policy makers are urged to amend patent laws to bring such inventions back into the scope of patentability. Alternatively, it is possible that the pharmaceutical industry could look to different innovation incentive models to fill this gap. Such different models could move away from thinking about protection based on a difficult-to-judge level of invention and monopoly pricing toward protection and pricing more transparently based on levels of investment. They might also open the door to more opportunities for not-for-profit drug discovery and development.
Looked at from these perspectives, it is quite clear that the use of AI not only offers the promise of improving drug discovery and development, hopefully making it less costly, risky and time consuming, but also very likely raises important questions for the pharmaceutical IP system. These include ‘nuts and bolts’ questions regarding the operation of the patent system and, for example, data exclusivity, as well as broader questions regarding whether the overall incentive structure they produce remains fit for purpose.
To explore these and other related questions, and to contribute to the wider debate taking place in intellectual property and ‘Access to Medicines’ circles, Medicines Law & Policy has published a new discussion paper: “Pharmaceutical patents and data exclusivity in an age of AI-driven drug discovery and development” (here). We cannot anticipate which of the hypothetical scenarios suggested in the paper is most likely to happen. It is possible that developments will head in an unanticipated direction altogether, certainly if humanity does indeed develop Artificial General Intelligence (AGI), at or above human-peer level, or even Artificial Super Intelligence (ASI), far above human-peer level. Nevertheless, it seems sensible to begin thinking about these issues now so that we reduce the likelihood of being taken by surprise in the future.
Christopher Garrison, MA LLM MA PhD, is a legal advisor with over 20 years of experience on intellectual property issues.