Hacker News new | ask | show | jobs
by dgemm 162 days ago
State of the art 7 months ago is good enough for a lot of use cases.
3 comments

That was my thought. At my work, 7 months can be less time than it takes to get a project organized around a high-level plan. It sounds like nothing.
I suppose this is a more forward looking post though. It's about whoever gets to whatever is awaiting us in the future 7 months before the other.

And I guess the idea is is that there is this extreme inflection point in utility somewhere that makes it so getting there first gives you some incredible economic edge.

It might not exist though. Like either utility plateaus and its bubble crash of the century time or it just keeps going up but without any specific point where you can differentiate something.

Feels like that inflection point possibility passed a while ago since these models are starting to plateau in performance.
The GP is talking about recursive self-improvement.

What yes, it's clear by now it's way beyond the capacity of those AIs, and the odds are pretty good it's impossible to a large extent (but some limited version of it may be possible).

What makes you think they plateaued?
Exactly - for an average person's task, how different is AI today vs 7 months ago?