Herb Simon’s pioneering AI work on scientific discovery focused primarily on automatically deriving laws from data. As the complexity of the phenomena under study increases, learning from data alone has limited possibilities. Our work has focused on capturing scientific knowledge about data and analytic processes to develop intelligent systems capable of hypothesis-driven discovery. New research in intelligent systems will advance scientific research, by making data collection more effective, data integration faster, and data analysis more scalable. A crucial requirement for these intelligent systems, which sets them apart from other AI applications, is the need to capture all forms of scientific knowledge for reasoning and placing new discoveries in the context of what is known.
"Intelligent Systems for Scientific Discovery"