So I'm surprised how little discussion I've seen on this feature so I thought I'd post a thread to get people's impressions. I'm quite impressed. I have 750+ photos and the AI has done a pretty good job pulling things out where they're the main focus of a picture. Hits: it has successfully pulled out things like "cat" (with a couple of exceptions I'll note), "sports field", "tree", "palm tree", "woodlands", "waterways", "car" (even managing to separate sedan, suv etc.) Misses: There are a few odd misses I've noticed. I was typing "po" for ""pool" (which the AI did very well with btw) and I noticed there was a classification for "pooches". These were actually both cat pics, though both had been recently shaved in a "Lion cut" style so perhaps were mistaken for poodles. A couple of odd things too that I can sort of see the logic if I squint. Like I have a pic of some jerky in a white wrapper which is identified as ice cream. I guess if you squint it looks like vanilla with chocolate topping. The other "crime" is incompleteness. An odd shaped pool left out, an unusual dragster car not identified etc. All in all shaping up well for a first pass I think. Post your best/worst/funniest matches.