The questions studied in Artificial Intelligence Group (AI) are of enormous practical and scientific importance and have proven to be quite difficult for conventional programming techniques. Human brains evolved to excel at tasks such as vision, motor control, speech, and language understanding, and are much better at these than artificial systems. The AI Group effort is exploring how computational models and techniques based on natural intelligence can prove useful in applications tasks. The ICSI project differs from most others in its emphasis on structured networks, strong methods that exploit scientific knowledge, and extensive interaction with other computer science techniques and theory.

In particular, the AI Group continues its long-term study of language, learning, and connectionist neural modeling. The scientific goal of this effort is to understand how people learn and use language. The applied goal is to develop systems that support human centered computing through natural language and other intelligent systems.

Natural language understanding is a core activity of the AI Group. Three main efforts comprise this work: FrameNet, a project to build and exploit a machine-readable lexicon with detailed semantic descriptions of a substantial portion of the English vocabulary; the Neural Theory of Language (NTL), which uses computational models and simulations of language and learning to answer basic questions about the production and use of natural language; and Language Communication with Automomous Systems (LCAS), which explores "full path" language understanding with advanced systems like robots and automomous vehicles.

Professor Jerry Feldman is the AI Group leader.

Read about specific projects of the AI Group.