Why a promising AI drug can still be dropped | AI Learning Guide

Notice that none of these four reasons is about whether the molecule works. Novelty makes the risk harder to price. A thin evidence base makes funders wait. A conventional competitor already further along raises the bar. And a cheap early design does not help if the expensive stages after it are unchanged. Together they mean an AI-originated candidate has to clear a higher bar than a conventional one to get the same funding. That is how a drug can be real, published, and still never reach a patient.

0:00 / 0:00

Auto-play

Why the bar is higher for an AI-originated candidate

Novelty raises perceived risk, which lowers the probability term in the expected return.
A thinner evidence base makes the risk harder to price, so funders may wait for more data.
A conventional competitor already further along forces the AI candidate to be clearly better to displace it.
A cheap early design does not help if the expensive later stages are unchanged.

Being dropped is not a verdict on the molecule. It is a verdict on the project's risk-adjusted return — and that verdict can be reached before any patient is ever involved.