[D] 3 Reasons Why We Are Far From Achieving Artificial General Intelligence

I just wrote this piece which proposes an introduction to 3 challenges facing current machine learning:

• out-of-distribution generalization
• compositionality
• conscious reasoning

It is mostly inspired by Yoshua Bengio's talk at NeurIPS 2019 with some personal inputs.

If you are working or just interested in one of these topics, I'd love to have your feedback!