Yeah I've spent way too much time reading this "guy's" posts here, Academia profile, etc. Huge waste of time. AI has managed to amplify a crank 100x. This is only going get worse.

I've seen for myself how much tunnel vision these models will get when collaborating scientifically/mathematically. When working around unfamiliar domains I have to do extensive grounding on my own. Curious to see how this changes over the next two years as the industry goes after scientific collaboration.