Bold claims, small patient count | Visual AI Learning Guide

Look at the two columns together, because the point is the contrast, not either side alone. On the left are the claims: faster discovery, molecules no chemist would have drawn, failures predicted before expensive testing. On the right is what has actually reached patients: a handful of candidates in early human trials, and no AI-originated drug yet in routine clinical use. Notice that the left column describes the design stage, while the right column counts completed journeys. Those are different things, and the distance between them is what we are going to explain.

0:00 / 0:00

Auto-play

The public claims about AI in drug discovery are broad. They include discovering new drug candidates in a fraction of the usual time, designing molecules that human chemists would not have thought of, predicting which candidates will fail before expensive testing begins, and eventually making drug development cheaper and more reliable.

The observable reality is narrower. A small number of drug candidates that originated from AI-driven design have entered human testing, mostly in early-stage trials. Entering a trial is not the same as succeeding in one, and succeeding in an early trial is not the same as becoming a medicine that doctors prescribe. As of the mid-2020s, no AI-originated drug had become a standard treatment in routine clinical use.

Both columns in the comparison are real. The claims describe what the tools can do at the design stage. The reality describes how few of those designs have completed the long path to a patient. The gap between the two columns is the puzzle this course sets out to explain.