| Abstract Scope |
Artificial intelligence (AI) offers the tantalizing possibility of accelerating materials discovery, optimization, and deployment. AI can standardize synthesis and characterization experiments, curate data and knowledge autonomously, and help automate tasks that are hazardous, repetitive, or typically error prone. AI can also support efficient data analysis and information extraction from large data streams typical of beam line experiments and high throughput computation. It is possible to analyze material microstructures rapidly using AI. When combined with physical laws and practical constraints, AI can generate insights from complex datasets, enhance the depth of analysis, and uncover subtle trends in the data. Physics-informed AI is an important step to make models more transparent and explainable. However, there are several challenges to the widespread use of AI in materials research, such as limited availability of relevant high-quality data with metadata, the need to quantify uncertainty, complex relationships, and sloppy use of AI. The talk will highlight our recent progress in this field and present a path to move from AI and automation to autonomous experiments. |