And we see things like Grokipedia has actually really struggled from the fact that by replacing humans with just AI, they’ve had lots of problems. One of the articles cited the Kremlin, right? And a volunteer would have got that immediately: wait a minute, this is an article on Russia, you’re citing the Kremlin — that’s obviously a biased source. And the AI just didn’t get that part. And there is something that through discussion and through consensus-building — that is a magical model that’s worked for a really long time. And the LLMs are just not there yet.