Discovered Materials offers $1M for AI solutions to 5 materials problems

In short: Discovered Materials is offering a total of $1 million for solutions to five long-standing materials science problems, and it wants AI systems to try.

What happened

Discovered Materials, a young San Francisco startup backed by Y Combinator, announced a prize challenge worth $1 million in total. The company says it will pay $200,000 for each solution to five open problems in materials science.

Materials science is the study of what things are made of and how they behave. Discovered Materials uses AI models, meaning computer programs trained on lots of data, to help find new substances that could be used in advanced computer chips.

The five problems include explaining “cuprate superconductivity,” where some materials can carry electricity with zero resistance at very low temperatures. Others focus on better ways to predict how materials behave at tiny scales, like what happens where silicon meets silicon dioxide, a pairing used in most semiconductors. One problem also targets improving “Density Functional Theory,” a common way scientists simulate how atoms and molecules are arranged (like a computer model of how LEGO pieces fit together).

The startup worked with university researchers to define the problems and how to check answers. It also tested today’s top AI systems by letting them work for days with internet access, and said they did not make much progress.

Why it matters

Better answers in materials science can affect everyday products, including faster electronics and improved batteries. The challenge is also meant to act like a scoreboard for AI in this area, so researchers can tell whether AI is actually getting better at explaining how real-world materials work.

Source: TechCrunch AI