AI-powered bioengineering research has been taking off.
- A new system, called ProtGPS, predicts how proteins sort themselves within a cell. It joins Nobel-winning AlphaFold in the world of next-generation structural biology tools.
- LinkedIn founder Reid Hoffman and Pulitzer Prize-winning author and cancer scientist Siddhartha Mukherjee launched Manas AI, their AI-powered drug discovery startup focusing on cancer.
- OpenAI announced GPT-4b micro, a model that designs proteins for longevity science—more on this story in next week’s newsletter.
Pharma and biotech startups like Manas AI and Paris-based Owkin are doubling down on what AI can do for the tedious process of drug discovery. An Owkin executive insists that AI can help their company double drug discovery success rates—from 10% to 20%.
Still, in the same interview I’ve linked above, the Owkin executive admits that, while AI gets its attention for “sexy” headlines about clinical and research applications, it’s the “boring stuff” of administrative automation where these tools will truly make a difference for health innovation.
This is a point we’ve heard before—and the types of medtech AI startups making the most headway in the industry corroborate the theory.
Yet, if even drug discovery executives believe AI is more ready to help with backend office work than hard science, what are we to make of the massive claims of what AI can do for drug discovery? Are they really just marketing and fundraising hype, as some critics claim?
Certainly, many of these companies have made tangible progress in their pipelines with the help of AI.
What’s in contention, it seems, is whether AI is truly pulling new drugs out of thin air. Or are we changing what we mean when we call something de novo design?
These discussions, to me, mirror the broader cultural discussion about what AI can truly create. Whether it’s art or drug design, we’re redefining how we understand novelty, authorship, and iteration.
At the same time, we must be mindful of our communication about AI and the credit that we give a still nascent technology. Progress in medtech depends on dreaming big and funding moonshots—let’s not get stuck in the weeds arguing over semantics.
