AI Enhanced Science

AI-Enabled Research and Scientific Workflows

We are developing systems and methodologies that use large language models (LLMs), multi-agent systems, and agentic AI to accelerate scientific and engineering research. Rather than limiting AI to isolated tasks such as classification or data analysis, our goal is to support larger portions of the research workflow while keeping the researcher responsible for scientific judgement and validation.

Our systems support literature discovery, research planning, data processing, analysis, visualization, and scientific reporting. For example, raw experimental inputs such as high-speed videos, images, sensor signals, and measurements can be processed by specialized agents to extract parameters, generate plots and regime maps, connect findings to prior literature, and produce a traceable first draft of a manuscript.

This work also contributes to our broader PAIOS (Personal AI Operating System) vision, where persistent AI capabilities, specialized agents, tools, and research memory work alongside the scientist.

We are also developing AI-assisted methods for experimental planning, data acquisition, and data reduction to help researchers work more efficiently and reproducibly. These approaches build on our earlier use of machine learning for focused scientific problems and extend across research, engineering analysis, scientific communication, and education.