总部位于旧金山的Y Combinator 2026年春季批次初创公司Discovered Materials宣布设立100万美元奖金池,专门用于奖励解决五个长期未决的材料科学问题 1。该挑战由联合发起人Felipe Jornada和联合创始人Advaith Sridhar共同推动,其中Jornada指出,攻克这些难题将直接影响电子制造、能源存储及未来材料的生产 1。Sridhar表示,此举意在为人工智能在材料科学领域的进步设定明确的“球门柱”(目标基准),以评估前沿模型在复杂物理领域的能力 1。
这五个被选定的开放性问题涵盖了多个核心科研领域,具体包括解释铜酸盐超导性的理论、预测硅与二氧化硅界面结构的数值方法、映射晶粒结构与热性能的数值方法、预测光照改变材料行为的理论,以及改进密度泛函理论(DFT)的数值方法 1。这些问题由来自斯坦福大学等顶尖高校的学者协助制定,并配备了明确的验证标准以确保评估的客观性 1。目前的测试结果显示,尽管现有的前沿AI模型已自主运行数天,但在解决上述问题上仍未取得重大突破 1。
Discovered Materials, a San Francisco-based startup from the Spring 2026 batch of Y Combinator, has launched a $1 million prize pool to reward solutions to five open problems in materials science 1. The initiative aims to establish new benchmarks for evaluating how well advanced AI models can handle complex physical and material domains by providing clear verification standards 1. Co-founder Advaith Sridhar described the challenge as setting "goalposts" for progress in AI-driven materials science 1.
The five specific problems targeted by the prize include developing a theory to explain cuprate superconductivity, creating numerical methods to predict the structure of silicon-silicon dioxide interfaces, mapping grain structures to thermal properties, formulating theories to predict how light alters material behavior, and improving numerical methods for density functional theory (DFT) 1. These challenges were defined with assistance from scholars at top universities, including Stanford University 1. According to co-founder Felipe Jornada, solving these issues would have direct impacts on electronics manufacturing, energy storage, and the production of future materials 1. Testing has shown that existing state-of-the-art AI models have not yet made significant autonomous progress on these tasks even after working for several days 1.
评论
还没有评论,欢迎留下第一条。