Abstract
How could machines learn as efficiently as humans and animals? How could machines learn to reason and plan? How could machines learn representations of percepts and action plans at multiple levels of abstraction, enabling them to reason, predict, and plan at multiple time horizons? This position paper proposes an architecture and training paradigms with which to construct autonomous intelligent agents. It combines concepts such as configurable predictive world model, behavior driven through intrinsic motivation, and hierarchical joint embedding architectures trained with self-supervised learning.
Meta's Chief AI Scientist Yann Lecun lays out his vision for what an architecture for generally intelligent agents might look like.
It probably would have gotten more engagement if someone else (e.g. Gwern posted it). I'm a low karma/unpopular account, so few of my posts get seen unless people go looking for new posts.
I rarely read new posts (I read particular posts on things I'm interested in and have alerts for some posters). So, it's not that surprising, I guess? I wouldn't have read this post myself had someone analogous to me posted it.
Maybe there's a way to increase discoverability of posts from low karma/unpopular users?