upvote
Not splitting hairs at all, but I recall a lot of mid-2030s estimates around that time period. The horizon has drawn in a lot in a few years.
reply
Yeah, I remember that last decade the prediction was the 2040s, but then LLMs happened and it became 2030s.
reply
Mid-2030s estimates aren't uncommon even this year.

And then there's this gem: https://www.dwarkesh.com/p/ege-tamay "AGI is still 30 years away — Ege Erdil & Tamay Besiroglu"

reply
I was _hoping that maybe_ 2050 and other software engineers were considering me a crazy. GPT-2 was out by then.
reply
As I said to someone else, this wasn't by major figures, this was in conversation with ordinary people. As I said to them, if you want to dismiss this you'd have to dismiss OPs prediction too, and probably cease making your own unless you can get a news-outlet to cover it
reply
Why do you believe that "OP" is making a prediction on his own?

There are some well-known experts saying this, like Paul Christiano, Daniel Kokotajlo, etc. Both are ex-OpenAI and are advocating pause/slowdown.

reply
That kind of feels like moving the goalposts. Nobody becomes Nobody of significance becomes Nobody of significance without others of significance making similar claims, all 3 of these falling under the banner of "Nobody".

But at this point its a matter of semantics, and wasn't really the point I was trying to make. And as I said to someone else, if the reason for making these predictions is to encourage a slowdown I'm much more sympathetic to them (since there a real reason to act of them)

reply