←back to thread

I don't think AGI is right around the corner

(www.dwarkesh.com)
337 points mooreds | 3 comments | 06 Jul 25 20:45 UTC | HN request time: 0.595s | source
Show context
vessenes ◴[06 Jul 25 21:52 UTC] No.44484424[source]▶
Good take from Dwarkesh. And I love hearing his updates on where he’s at. In brief - we need some sort of adaptive learning; he doesn’t see signs of it.

My guess is that frontier labs think that long context is going to solve this: if you had a quality 10mm token context that would be enough to freeze an agent at a great internal state and still do a lot.

Right now the long context models have highly variable quality across their windows.

But to reframe: will we have 10mm token useful context windows in 2 years? That seems very possible.

replies(4): >>44484512 #>>44485388 #>>44486146 #>>44487909 #
1. nicoburns ◴[07 Jul 25 00:14 UTC] No.44485388[source]▶
How long is "long"? Real humans have context windows measured in decades of realtime multimodal input.
replies(2): >>44487895 #>>44489678 #
2. MarcelOlsz ◴[07 Jul 25 08:15 UTC] No.44487895[source]▶
Speak for yourself. I can barely remember what I did yesterday.
3. vessenes ◴[07 Jul 25 12:29 UTC] No.44489678[source]▶
I think there’s a good clue here to what may work for frontier models — you definitely do not remember everything about a random day 15 years ago. By the same token, you almost certainly remember some things about a day much longer ago than that, if something significant happened. So, you have some compression / lossy memory working that lets you not just be a tabula rasa about anything older than [your brain’s memory capacity].

Some architectures try to model this infinite, but lossy, horizon with functions that are amenable as a pass on the input context. So far none of them seem to beat the good old attention head, though.