Chemistry

AI catalyst predictions fail when experimental conditions vary across labs

How the science connects

Artificial intelli…CatalysisReproducibility

AI Insight

Four independent laboratories conducted experiments to test AI model predictions for catalysts that convert carbon dioxide into useful fuels. The study revealed that experimental differences between labs can significantly affect results, undermining the reliability of AI predictions for catalyst performance. This highlights that AI models for catalyst discovery are highly dependent on the quality and consistency of the experimental data used to train them.


This research exposes a critical challenge in using AI to accelerate the development of catalysts for carbon dioxide conversion, a key technology for reducing greenhouse gas emissions and producing sustainable fuels. The findings suggest that standardized experimental protocols across laboratories are essential before AI can reliably guide catalyst selection and development.


To turn abundant carbon dioxide into valuable fuel, we need a fast and efficient way to determine which catalysts work best over the longest time. AI models have the potential to help guide catalyst selection, but as with internet chatbots, AI models are only as good as the data you put into them.

Source: Four labs find experimental differences can undermine AI catalyst predictions