Goldsea set out to identify the most consequential materials breakthroughs enabled by artificial intelligence by posing a simple prompt to two widely used chat systems. The responses paint a picture of discovery at a pace that now exceeds the speed at which new compounds can be engineered into mass-market products. Both lists elevate Google DeepMind’s GNoME project, which applied artificial intelligence to propose hundreds of thousands of thermodynamically stable inorganic crystals. One tally highlights 381,000 structures overall, while the other singles out 52,000 graphene-like layered compounds with potential in electronics, superconductivity, and quantum devices.
Energy applications dominate. The chat systems emphasize artificial intelligence guided searches that surfaced high-conductivity lithium solid electrolytes, viewed as critical to safer, denser all-solid-state batteries. They also cite accelerated design of new perovskite and organic semiconductors that could quicken the path to cheaper, printable solar panels. On the climate front, the lists point to artificial intelligence workflows that screened vast families of metal-organic frameworks to pinpoint top performers for carbon dioxide capture and gas separations, as well as catalysts for green hydrogen and electrochemical ammonia synthesis. One entry notes GNoME’s identification of 528 lithium-ion conductors, and another credits MIT’s CRESt platform with an eight-element electrode that improved direct formate fuel cell power density per dollar by 9.3 times.
Computing and quantum materials feature prominently. NIST’s CAMEO system is credited with autonomously discovering GST467, a Ge-Sb-Te phase-change alloy with twice the optical contrast of conventional materials, promising gains in data storage and neuromorphic or photonic computing. Several entries spotlight two-dimensional magnets, altermagnets, and kagome-lattice candidates uncovered by artificial intelligence tools for future spintronics and quantum technologies, along with a Johns Hopkins APL superconductor discovered through predictive modeling. The lists also include polymer advances such as a newly discovered polymorph and mixed ion-electron conductors, a multiple principal element alloy (FeNiCrCoCu) designed via explainable artificial intelligence, and stronger, more flexible rubber-like polymers from a hybrid artificial intelligence-human design workflow.
Not every breakthrough targets chips or chemistry labs. One selection highlights super-cool, radiative “ultra-white” paints engineered with artificial intelligence that can reflect more than 95 percent of sunlight, passively cooling buildings and reducing air-conditioning loads. Taken together, the ChatGPT and Grok responses underscore how artificial intelligence is expanding the palette of viable materials across electronics, energy storage, climate mitigation, and quantum science, even as large-scale validation and manufacturing remain the next hurdles.
