6 Comments
User's avatar
Theophrastvs's avatar

Good for you for promoting your friends endeavors! (But currently what passes as "A.I." is none of those things. It's a probability determined (hence) Markov chain built on a very large, mostly stolen, mass of data, that's all)

Dogfriend's avatar

It’s a digital parrot.

Ryan's avatar

Very energy and water hungry MadLib

Mark Frauenfelder's avatar

Not sure if you read the essay, but Kevin is saying that LLMs are very likely the wrong tool to be using.

Theophrastvs's avatar

While I did read his "Without a theory of intelligence" (in which he states 'scientists have no theory of intelligence'; which itself would shock (*shock*) some evolutionary biologists i know, who have theories in abundance) it was your second paragraph above, presuming great revolutionary strides in novel awareness which an engine reshuffling nothing but our already catalogued knowledge provided some suffering doubts. May i be proven wrong. Yet having cobbled together a couple of 'A.I.'s myself and seen their clear limitations, i'll be shocked *shocked*. They really are stochastic "mad libs".

Leo's avatar

AI is a faulty tool. If you point it at your own mind you will get faulty results.