The learn phase is what separates a loop from a toolchain. The structured measurement travels back to the model, which updates its internal parameters so that its predictions shift. The next molecule it proposes is therefore different from the one it would have proposed before the experiment. A toolchain passes work from one tool to the next but each tool stays fixed; a loop lets each result change the next design automatically.
Can AI Run a Drug Discovery Lab on Its Own?
Closing the Loop: Self-Driving Labs
The Result Changes the Next Design
3 / 3
Now the record travels back to where the cycle started. It flows into the model, and the model does not simply store it. The numbers inside the model shift, so its predictions move.
0:00 / 0:00