This AI stuff is wild. I tried looking people up, like Art "There is No Significant Oil In US Shale" and it did a reasonable job of being...polite. Lacked detail though, particular when he hosed things up. So I looked up myself and that was interesting as well, it found more by just making a small change, but it reasonably quantified and described published works, area of expertise, and contributions to science and whatnot. More than I expected it to know. I've used it for some coding recently, and that was less effective, but still I learned from some of what it laid out.

Not sure I'll take its word for collapse though, considering how differently it has been characterized by the true beleivers.

It knew that David White at the USGS was predicting peak oils as far back as 1919. That isn't a well known one, but it knew. I'll have to poke around on deep research issues and see if it knows that, then ask it to explain why so many folks got it wrong spanning a century and see if it figures out the trick in the question.

It has already made a mistake on hydraulic fracturing, claiming Hubbert didn't know about it in 1956, which he certainly did. It didn't seem to know the timing of when horizontal drilling began either, assigning to a post 1970 time frame as to why Hubbert was wrong.