Může umělá inteligence vyvíjet nové udržitelné materiály ?

Vývoj nových materiálů je klíčový pro pokrok technologií a snižování naší ekologické stopy. Umělá inteligence se uplatňuje v této výzvě s potenciálem objevovat nové materiály s unikátními vlastnostmi. Analýzou obrovského množství dat o složení a vlastnostech materiálů může AI předpovídat chování nových materiálů a navrhovat kombinace, které dosud nebyly vyzkoušeny. To by mohlo vést k průlomům v oblastech jako je ukládání energie, stavebnictví a elektronika. Využití AI ve vědě o materiálech také slibuje urychlit objevovací proces, čímž se sníží čas a náklady spojené s tradičními metodami pokus-omyl.

Background

The development of new materials is crucial for advancing technologies and reducing our environmental footprint. AI is being applied to this challenge, with the potential to discover novel materials with unique properties. By analyzing vast amounts of data on material composition and properties, AI can predict the behavior of new materials and suggest combinations that have not been tried before. This could lead to breakthroughs in fields such as energy storage, construction, and electronics. The use of AI in material science also promises to accelerate the discovery process, reducing the time and cost associated with traditional trial-and-error methods. As the world seeks more sustainable solutions, the role of AI in material development is becoming increasingly important.

AI is already contributing to the discovery of new sustainable materials by accelerating simulations and screening vast chemical spaces, for example using generative models to propose candidate molecules and density-functional theory to evaluate stability and performance. Recent systems like GNoME, MatterGen and AlphaTensor have identified thousands of stable inorganic structures and even novel superconductors with reduced trial-and-error, while robotics-driven labs such as those at DeepMind and Carnegie Mellon are closing the loop by autonomously synthesizing and characterizing promising candidates. Although human expertise remains critical for setting objectives and interpreting results, AI is demonstrably able to propose viable new materials faster than traditional methods, cutting design-to-discovery timelines from years to months.

— Enriched May 12, 2026 · Source: DeepMind