Convolutionary, Evolutionary, Revolutionary: What’s next for Bodies, Brains and AI?

Peter Stratton

TL;DR: Limitations of current AI are generally recognized, but fewer people are aware that we understand enough about the brain to immediately offer novel AI formulations.

Keywords: spiking neural networks, self organization, feedback, oscillations, predictive models, dynamic correlations, stochastic sampling, RL, embodiment

Abstract: In recent years we have made significant progress identifying computational principles that underlie neural function. While not yet complete, we have sufficient evidence that a synthesis of these ideas could result in an understanding of how neural computation emerges from a combination of innate dynamics and plasticity, and which could potentially be used to construct new AI technologies with unique capabilities. I discuss the relevant principles, the advantages they have for computation, and how they can benefit AI. Limitations of current AI are generally recognized, but fewer people are aware that we understand enough about the brain to immediately offer novel AI formulations.