IntelligenceAtScale

ThoughtStorms Wiki

A debate I occasionally find myself in on Twitter is talking about the problem of BigData / MachineLearning / ArtificialIntelligence which is edging towards an assertion that these are wrong because the data is "too big".

Eg. here I wrote

One of the questions that starts arising is, once we start talking about AI/ML as "centralizing" huge amounts of data, ALL knowledge is actually a kind of "data centralization" and summary. Even human academics reading a lot of books in a big library. \1

We have to be careful of any theory of what's "wrong" with AI/ML that is actually a theory that says that "intelligence" and "learning" are wrongs in themselves.

Because what follows from that?

But it's actually a serious and interesting question. Same as IsMassNatural.

You could undoubtedly make some kind of IvanIllich-like argument that the gathering of data and production of knowledge is a good thing at one scale but becomes harmful and damaging at larger scale.