My point was to reserve the generic term for something that's really intelligent, and thus deserving of rights an intelligent being should have, rather than what things really are. 'Machine Learning', 'Neural Nets', 'Magic black box Chinese Room simulations' (with millions to billions of parameter weights).

The public wants to hear the unspoken word General in the unqualified statement Artificial Intelligence. They correctly grammatically assume that this should refer to some synthetic entity created in a lab which has the ability to think.

For the LLMs that are popular today, I prefer the more truthful phrases I've heard. 'Transformers' (transform an input prompt into a related output) or more loosely 'codecs' (mostly decoders) which synthesize an output based on a prompt and an internal cache of structures.

I personally consider the compression of ideas into more generalized structures to be a big component of learning for humans too. However it isn't sufficient to simply learn, understanding is required. The output I've seen doesn't cross from tool to person.