It briefly discusses DALL-E and imagine models near the end, which seems to paper over a big part research in AI right now, but that also wasn't the point of th

It briefly discusses DALL-E and imagine models near the end, which seems to paper over a big part research in AI right now, but that also wasn't the point of the article, which is fair. What I find interesting is that this reminds me how AI is being trained on a very narrow set of human inputs. Images, text, possibly some video, but there's a huge amount of data governing human behavior that is none of those things. Facial expressions, body language, tone, diet, environment, etc. etc. etc. I'm sure it's coming, but it feels a long way off.