research

I want to understand intelligence well enough to build it, and well enough that we are not confused about what we built.

Right now that means learning dynamics: how a high-dimensional system forms a representation, how that representation moves with data and time, and what any of this has to do with the brain. Depending on the question I reach for neural computation, statistical mechanics, or deep learning theory.

The longer hope is a system that can think about itself and about the world, with internals we can actually follow.