Eyal Amir

Ph.D. Stanford University, 2001

Research Statement

I am interested in artificial intelligence (AI), with an emphasis on autonomous systems that learn and make use of explicit knowledge to act in large, complex, partially observed worlds. This involves using principled, mathematical tools from logic, probability theory and decision theory, together with cognitive science concepts and techniques from the theory of computation.

Technically, I apply mathematical and algorithmic tools to study the representation of knowledge (e.g., using logic and probability), find efficient automated reasoning procedures for such knowledge, design autonomous-agent architectures that embody these algorithms, and present learning algorithms for such agents. Application areas that attracted me so far include robotics, virtual interactive worlds, vision, information retrieval, and text understanding.

More broadly, my work combines theoretical and applied AI. I believe that the two are best pursued together. Insights that we gain from theoretical research bring about ideas for applications that otherwise we would not imagine. Applied ideas ground theory and raise theoretical questions that would not be considered as problems otherwise or whose importance would not be recognized.

The driving force behind my research is the goal of human-level AI, and I believe that this goal is best pursued using the rapid development of important intermediate technologies. I believe that investigations into human intelligence and the development of techniques and tools that humans do not have will lead to this goal.

In the News