My take is HN is changing to adapt to current changes to the hard question: What separates intellectually interesting AI from uninteresting AI, right now?

My take is HN is changing to adapt to current changes to the hard question: What separates intellectually interesting AI from uninteresting AI, right now? Thirty years ago, Baysian generated poetry was intellectually interesting enough that Paul Graham put some in one of his programming books…yes, it made it through an actual professional editor and the publisher’s money was spent printing the poetry on a few pages instead of printing something else. A ten years ago, code-completion was intellectually interesting. A five years ago, images from text was. Today, none of them generally are except occasionally at the leading edge. Or to put it another way, there’s an ice machine on the airplane taking you to Dallas Fort Worth and we just take it for granted.