Gary Marcus Demands Evidence For AI Progress And Goalpost Claims
Reactions from ranked influencers
2 posts
my feed is filled with cope claims that goalposts have been moved, without any specifics whatsoever and claims that hard problems have solved, without any specifics whatsoever y’all want to disagree with me, fine. but come to me with *evidence*. (see post below for a sampling of my own receipts from last 25 years)
some of my own many receipts:
For the record, I never said that AGI would never come! - I have said it would very likely come this century (see debate with Grady Booch on my substack) - i have said that it might well come next decade - I said pure LLMs wouldn’t get there (they didn’t) - I said that adding in (neuro)symbolic components would help (they have helped; harnesses, tools, loops, etc) - I said that neurosymbolic components would be necessary but not sufficient (seems to be true, see my 2020 Next Decade in Arxiv) - I gave criteria in my essay Dear Elon Musk and my bet with Miles Brundage (both in my substack) Note also that so far all of these claims have either proved correct or pending but looking promising. And that receipts abound.