Hacker News Reader

ML is, obviously, AI. But not all AI is ML.

The reason the term "machine learning" was even invented was because it was one of the AI winters and an euphemism was needed because "AI" was more of a swearword than a buzzword.

Every AI generation has been useful. No AI generation has been AI, including this one.

"now"? The appliance industry pretty much hit the "label everything AI" button...again...within months of ChatGPT taking off.

Last time around was when "fuzzy logic" came out, I think?

I guess the appliance industry has dibs on the name AI anyway.

AI used to be called ML

which used to be called Statistics

which used to be called “math” or maybe “applied science.”

Obviously the underlying tech and research changed along the way… but not as much as it would seem. We’re still doing matrix operations and gradient descent and softmax, all of which has been around for a while.

This is not true because many older AI disciplines were not machine learning. A lot of work was put into tree search, logic programming, etc that we don’t count as ML. Older NLP work had little or no machine learning either.