“Scientists looking to remove carbon dioxide (CO2) from the air cleanly and cheaply have long been interested in metal-organic frameworks, or MOFs: gigantic, sponge-like molecules that can be precisely engineered to capture the gas and then release it on command.

Made of metal ions held together by compounds containing carbon, MOFs come in a dizzying array of structures, each with its own distinct properties. A MOF capable of absorbing CO2 at a humid sea-level location, for example, will have a different structure from one that can operate in a dry, high-altitude climate. Sorting through the billions of possibilities to find the right MOF for the job is an almost impossible task for a human chemist. It is, however, a perfect task for an artificial-intelligence (AI) model.

One startup attempting to build such a model is CuspAI. It has used the Llama family of open-source large language models, produced by Meta, a tech giant, as the basis for its model, which boasts between 5bn and 8bn parameters and has been fine-tuned on a vast quantity of materials-science data, from quantum-mechanical simulations to academic papers describing fabrication methods. CuspAI’s goal isn’t simply to find a good MOF, but to build a system that can spit out the right one for any environmental conditions—and, from there, to demonstrate that AI can be used to tackle any problem in materials science. Better batteries, cleaner bioplastics, more powerful semiconductors and, potentially, even room-temperature superconductors might soon be up for grabs.”

From The Economist.