Initiatives are already under way to avoid ill-considered moves concerning the impact of artificial intelligence (AI) on drug patenting (see L. Heuer Nature 558, 519; 2018).
Heuer mentions some of the issues. For example, he foresees problems over whether to designate the algorithm or its programmer as the inventor, and whether a drug discovered through machine-learning methods would be patentable.
In the United States, at least, some of these issues are currently clear. For example, US patent law states that “a person shall be entitled to a patent”, and an algorithm is not a person. It also states that “patentability shall not be negated by the manner in which the invention was made”. More generally, it is insufficient to assert that just because an AI could arrive at a particular solution, then that solution must be obvious.
However, a serious problem for pharmaceutical companies is that, according to US law, only people can make the inventive step in patents. In practice, it is likely that algorithms are making many of those steps, raising questions about the validity of these patents in the United States. We welcome efforts to arrive at a consensus over such dilemmas by the robotics research community, intellectual-property professionals, the European Commission and the European Patent Office.
Nature 560, 307 (2018)