3 ms·
> But I think that’s generations away at best. I'm not sure whether you mean human generations or LLM generations, but I think it's the latter. In that case, I
by idopmstuff 2y ago
> But I think that’s generations away at best.
I'm not sure whether you mean human generations or LLM generations, but I think it's the latter. In that case, I agree with you, but also that doesn't seem to put you particularly far off from OP, who didn't provide specific timelines but also seems to be indicating that the elimination of most engineers is still a little ways away. Since we're seeing a new generation of LLMs every 1-2 years, would you agree that in ~10 years at the outside, AI will be able to do the things that would cause you to gladly retire?
- simianparrot 2y agoI mean human generations because to do system architecture, design and development well you need something that can at least match an average human brain in reasoning, logic and learning plasticity. I don’t think that’s impossible but I think we’re quite a few human generations away from that. And scaling LLM’s is not the solution to that problem; an LLM is just a small but important part of it.
- munk-a 2y agoI'd be cautious as describing anything in tech as human generations away because we're only about a single human generation into a lot of this industry existing.
- simianparrot 2y agoI understand the hesitancy to do so, but the thing is, what's changed with computers since the inception of the technology is speed and size. The fundamentals are still the same. Programming today is a lot more convenient than programming via punch cards, but it's a matter of abstractions + scale + speed. Given this I'm pretty confident when I say that, but obviously not 100%.