I'm also not sure why you'd think that, Apple's been at the forefront of "AI" for years now, running models locally and optimizing their CPUs for local workloads to e.g. identify people, places and pets (much appreciated lmao), create slideshows, and subtly improve photo's made on the device.
The photo organization is nice but that being said, if you try to use the on device Apple Foundation models you quickly find it is totally useless.