Tom Griffiths at Edge.org:
I work on computational models of cognition, which means that I’m interested in understanding how people do the amazing things that we do, like learning from small amounts of data, figuring out causal relationships, identifying languages—things that computers have traditionally found hard to do. The way that I think about motivating that kind of research is in terms of making computers better at solving those kinds of problems.
Recently, I’ve also been thinking about a different way in which that’s a relevant enterprise. With all of the successes of AI over the last few years, we’ve got good models of things like images and text, but what we’re missing are good models of people. If we look at the kinds of AI systems that are being built and the kinds of data that people want to understand, often those data have to do with human behavior. We're trying to understand why people do what they do and what the cognitive processes are that underlie the data we find in the world that are a consequence of human behavior.
This enterprise is important for a couple of reasons. It gives us the tools to make sense of these data that are becoming an increasingly important part of our lives. Also, having good models of how people think and behave is relevant to helping AI systems better understand what people want.
My approach is to try and understand the computational structure of the problems that people have to solve. If we’re trying to understand how people, say, learn a new causal relationship, how do we formalize that? How do we turn that into a math problem? That’s the kind of thing we can imagine getting a computer to solve.