BioTechBio
This is a post by Nathan Benaich worth reading for anyone interested in hashtag#AI and hashtag#drugdiscovery. The important TL;DR is that there is a lot to be excited about in spite of the recent failures of "AI-first" biotechs.
My own summary would be that (the hyperbolic Boger-Bong anti-techbio manifesto notwithstanding) AI is already having a tremendous impact on early drug discovery but it is best sandwiched between 2 layers of bio: TechBio should be BioTechBio, begin with biology, apply AI where it can make a difference, wrap up with biology. The middle layer will be expanding over time.
As an add-on to Nathan's post, as exciting as AI is to the early phases of drug discovery, the proof of the pudding is in the clinic. And AI has some major potential there as well. A drug candidate takes on a much larger role when pre-clinical and clinical studies are well designed and quality conversations are happening with the regulator, considerations that are often forgotten by the AI-first pioneers.
As a last point, end-to-end AI drug discovery and development should not be the guiding principle of any BioTechBio. Apply AI where it matters with the tools available today with an eye firmly on patient needs.