Researchers at the Ames Laboratory built an AI workflow called DuctGPT — a physics-trained model tasked with solving a real materials science problem: discovering rare-earth-free permanent magnets. These magnets are critical for EVs, wind turbines, and electronics, but current designs depend heavily on rare-earth elements that are expensive and geopolitically fragile to source.
What makes this stand out isn't just that the AI proposed new material candidates — it's that DuctGPT actually understood the underlying physics, inventing new materials while also factoring in production costs and sourcing constraints. That's a meaningful step beyond pattern-matching: it's AI reasoning through real engineering trade-offs, not just generating plausible-looking outputs.
This is what "AI for science" looks like when it moves past the hype cycle — not a chatbot demo, but a tool shortening R&D cycles on problems that matter for supply chains and sustainability.
لا توجد تعليقات بعد. كن أول من يعلّق!