Original Source Here
Deep learning is not enough.
You have the pieces of the puzzle in your article: causality, embodied AI, action and perception. Judea Pearl and al. have proven that observation (=data, no matter how many data) alone cannot infer causal models. Intervention is needed to test the world, find the arrow of causality and debunk counfounders behind mere correlation. This is why RCT is one of science’s gold standards. And this is why intelligence needs a body to perform independent/free/random intervention. You can inflate GPT-1000 or WuDao-1million as much as you want, and not get an inch closer to AGI.
Trending AI/ML Article Identified & Digested via Granola by Ramsey Elbasheer; a Machine-Driven RSS Bot