The AI revolution in chemistry has yet to happen due to a lack of accurate and accessible training data for machine-learning systems.
Researchers need more experimental and simulated data, including historical data and unsuccessful experiments, to establish comprehensive training data sets for AI tools in chemistry.