The coevolution of medicine and artificial intelligence

A couple recent developments evidence the synergistic relationship between medicine and AI.

In the category of medicine spurring innovation in AI, David Duvenaud and his collaborators received the 2018 “best paper” award at the prestigious Neural Information Processing Systems conference. Duvenaud hit upon a major shortcoming in AI as he was working on a project involving continuous medical data. He and his collaborators are now replacing the discrete layers of neural networks with calculus equations, a design they call an “ODE solver” (ordinary differential equations.” The work is being heralded as a major shake-up in AI, comparable to the introduction of GANs.

In the category of AI spurring innovation in medicine, the EU and a group of national funders have laid out a plan for open access to publicly funded research. Although there are many good reasons for it, a major driver is the need to support artificial intelligence. The overwhelming glut of scientific literature, along with its dubious quality, is driving the need for technology solutions. When existing business interests and culture stand in opposition to this need, AI may represent the straw that breaks the camel’s back.

This type of synergy and mutually reinforcing progress is why it is so difficult to project developments using one field in isolation. It’s also why the future may arrive much more rapidly than we might otherwise expect.