TL;DR
The first drug whose disease target and molecule were both picked by AI is now in the last and largest stage of human testing. Rentosertib, a daily pill for a fatal lung-scarring disease, began a 320-person, year-long Phase III trial in September. No AI-discovered drug has been approved anywhere, so this trial is the field's first real exam: a win would show that AI can find targets that beat a placebo, and a loss would suggest AI mostly makes discovery faster without making drugs better. Meanwhile money is pouring into the slow half of medicine, the clinical trial, through AI agents that run trial logistics and "digital twins" that might one day stand in for some placebo patients. If those pieces come together, the decade-plus it takes to bring a drug to market could start to shrink. If they do not, AI drugs will fail at the same rates as every other kind.
Below the line, for paid subscribers: The full story: where it stands, how we got here and what's dated next; our Odds board on 5 milestones, starting with "ARPA-H funds a platform for AI-run "phaseless" trials by 2027"; our labeled speculation on what it adds up to; and every source.