Why the gap is worth explaining | AI Learning Guide

The easy answer is that AI just does not work here, and that answer falls apart quickly. If the tools were useless, their molecules would never have interested chemists or reached human testing, and some have. So the gap has to come from somewhere else. Look at the list: behaving well in a real body, surviving years of human testing, attracting funding, convincing regulators, convincing doctors. Each of those is a separate filter, and each one removes most of what enters it. Designing a good molecule gets you past the first gate, not the rest.

0:00 / 0:00

Auto-play

The explanation that does not hold up

If AI were simply ineffective at drug discovery, its molecules would not be interesting to chemists and would not reach human testing at all. Some have. So the gap is not explained by the tools failing to work.

Filters a candidate must pass after design

Behaving well in a living body, not just in a model
Surviving years of testing in human volunteers and patients
Attracting enough funding to be developed to the end
Convincing regulators that the evidence supports approval
Convincing doctors and patients to actually use it

The question is not whether AI can design molecules. It is why designing them well has not yet changed what doctors can prescribe.