“Untold numbers of promising small molecule
treatments remain beyond
AI’s current reach…. To improve predictive power without unduly
sacrificing speed, there is one strategy that readily stands out — the
use of molecular modeling to overcome the data problem of training AI
models.”