Editorialized; original paper title:

(PCA) Principal component analysis of dendrite structures

They combined with persistent homology (PH), made it "interpretable". I'd say it's way more advanced than AI!

"principal component analysis (PCA), a machine learning technique"

More like: A old technique sometimes used in machine learning

The way articles are going they will be saying that a robot that punches buttons on a calculator [0] is artificial intelligence....

[0] https://www.youtube.com/watch?v=q-ZSxndNLPY

WOW PCA. That is some old ML right here. Last time I did PCA was for flows in an undergrad paper. It's been 15 years!

It is still pretty commonly used by statisticians, but framing it as "AI reveals" seems to be a stretch.

How about Latent Dirichlet Allocation? We used it to do sentiment analysis a little more than decade ago hehe

PCA was introduced in 1901. Who knew Karl Pearson was actually doing AI!

The image processing in a macaques brain has been doing PCA since way longer. Atleast according to [1], around 1hr36m in.

https://www.youtube.com/watch?v=Uj6skZIxPuI

AI?!

This is just old school ML algorithm.

Well, AI is how we call machine learning these days. The deep neural networks we are using today are not exactly a new thing either (1960s), it is mostly just scaled up because we now have the hardware to do it.

AI is a vague term, ranging from simple game playing engines to humanlike sci-fi robots and godlike superintelligences. But now, most of the times, it just means machine learning.

ML is a subset of ML

It is not a proper subset, though.

I meant ML is a subset of AI ...

It actually predates computing machines by about half a decade.

Sorry, half a century.