Yep! Honestly it was just a random rabbit hole - I recently read Lee Smolin's Time Reborn (highly recommend btw, super fascinating read) and was curious as to what his more recent work was about, which lead me to come across The Autodidactic Universe paper. With the AI hype train full steam ahead, the paper felt newly relevant, especially as it seems that we're starting to hit plateaus for model intelligence and looking to other areas (e.g. world models) for further advancement in the field
