FDA launches 2026 Accelerated AI Pathway Pilot to evaluate AI-generated evidence in drug submissions

What people don’t realize about the recent phenomenal success of the metabolic drugs is that ALL of them were designed/created *pre* the current AI boom. We have not even started to scratch the surface of what is possible with these new tools. But the promises are unbelievable.

Source: https://intuitionlabs.ai/articles/fda-accelerated-ai-pathway-pilot-phase-1

@kimmonismus most of those 200 entered trials in the last 2-3 years, fda approval takes a decade minimum, so zero approvals at this stage is kind of expected

@netrunner_btc yeah, but it shows at the same time that FDA needs so speed up their approval rate

@kimmonismus We are about to witness the end of diseases

@tunguz I feel there will be an important bottle neck somewhere. For example we might discover a whole bunch of drug candidates, peptide strings and so on. But testing them all and ensuring quality, getting it to production and so on will also be something that needs to be solved

@Creatus_AI and i couldnt be happier

@kimmonismus 200+ clinical candidates without approvals suggests discovery throughput has gone crazy

@kimmonismus 0/200 isn't a fail signal yet — base rate is 90%+ for all clinical-stage drugs, and AI-led design is mostly post-2020 work. Phase-3 takes 6-10 yrs. Real comparison: time-to-phase-3 for AI drugs vs traditional. FDA pilot is the right move — regulation is the bottleneck.

This is exactly why I’m so optimistic.

The drugs didn’t fail.

The science started moving faster than the rulebook.

For people living with serious conditions, that matters. AI may compress years of discovery into months, surface targets humans would miss, and turn biology into something searchable.

The FDA catching up is not bearish.

It means the future finally forced its way onto the desk.

@kimmonismus oh how often have i quoted this already lmao

@kimmonismus 200 candidates in trials, zero approved. That has little to do with AI getting ahead of regulators and nothing to do with framework failure. Most entered trials one to three years ago. The FDA pilot is adaptation, not confusion.

@kimmonismus 200 drugs in trials. the FDA is still figuring out how to grade the test.

@kimmonismus This is the pattern everywhere.

The product gets announced. The investment gets celebrated. The trials begin. The framework for accountability arrives later.

Maybe that is acceptable for a chatbot feature.

It is a much stranger way to handle medicine.

@kimmonismus Some combination of Ai + quantum will solve this, eventually. End to end human simulation is needed, but it's a real challenge that may be a decade away.

@kimmonismus @netrunner_btc But isn't the time they set is for ensuring the drugs has no side effects?

@kimmonismus the 10 companies fda picked for the pilot are the story, not the program. whoever shapes how ai-derived evidence gets evaluated sets the regulatory moat for the next decade. ai pharma policy is being written in a closed room

@kimmonismus This is the reason why I don't believe in Kurzweil's LEV theory. Yes, the medicine is there but it's not being released due to all these long ass test periods.

200+ in trials, zero approved — the gap between R&D velocity and regulatory infrastructure has never been wider. The pilot is a necessary first step, but the real signal will be how the FDA defines "AI-generated evidence" as distinct from traditional statistical analysis.

@kimmonismus @netrunner_btc They really do. Instead of 10 years and a billion dollars to bring a new drug to market, it should be a year and 100 million. Or something like that.