Skip to content
Learn Motion
ExploreHow it worksMembership
Log in
Learn Motion

Can AI Discover New Drugs? A High-Level Overview

1Why Drug Discovery Is Hard, and Where AI Fits2How AI Learns From Molecules and Proteins3Finding and Validating a Biological Target4Designing Molecules: Generative AI and Virtual Screening5From Hit to Lead: Optimizing Properties With AI6What AI Still Cannot Do7Judging the Claims: Real Successes, Failures, and Open Questions
Why Drug Discovery Is Hard, and Where AI Fits

Why Almost Everything Fails

2 / 4
Follow the funnel from the wide top to the narrow bottom. The width at each level is the number of compounds still alive. Most of the narrowing happens near the top, where compounds are cheap to make and cheap to discard. But look at the last few levels — the survivors there are few, and each one represents years of investment. When a compound fails at that depth, the entire accumulated cost is written off. That is why the field cares less about making the early stages faster and more about making the late predictions more trustworthy.
0:00 / 0:00

The pipeline diagram makes the process look orderly, but the numbers behind it are brutal. For every medicine that reaches patients, thousands of compounds are made and tested, and the great majority are discarded. The narrowing is not uniform: most compounds are eliminated early and cheaply, during hit finding and lead optimization, but a meaningful fraction survives all the way to human trials and then fails there, where each failure costs far more.

This pattern is called attrition, meaning the rate at which candidates drop out of the pipeline. Attrition is the central economic fact of drug discovery. Because the cost of the late failures must be carried by the few successes, the average cost of bringing one approved drug to market is commonly estimated in the range of one to several billion dollars, and the timeline from first target work to approval typically runs ten to fifteen years.

The practical consequence is that reducing late-stage attrition is worth far more than reducing early-stage cost. A method that eliminates a thousand compounds in the laboratory saves relatively little; a method that correctly predicts which of two candidates will fail in phase 3 saves an enormous amount. This asymmetry is the reason the field pays so much attention to any tool that might improve the reliability of early predictions.

Previous2 / 4Next

Learn Motion

Generate a course. Learn it properly.

Operated by Wuhan Daoyin Technology Co., Ltd.

Contact: [email protected]
Privacy PolicyTerms of Service

© 2026 Learn Motion