Is AI Progress Hitting a Wall? No

[jaegerpicker]

I think it is slowing down, almost undeniably slowing down. Hitting a wall? I don't think that's fair but slowing down for sure. Which follows the tech hype process, an engineer(s) make great progress on a problem, an "idea" guy hypes it to the moon and sells it for as much money as possible by promising the moon, and then the tech "fails" to deliver the moon. Even the tech was never meant to deliver said moon.

[mberlove]

It's easy to get negative about a seemingly narrow technology. It's easy to want to compensate by becoming overly optimistic. I have trouble fluctuating between these since it's also easy to make convincing arguments for both -- point to a swell of articles for progress, point to a stagnation as lack of progress -- but all technologies go through this.

IMHO it's premature to have a strong conclusion on this, and way too easy to let feelings, pro or con, influence our projections.

[beardyw]

I think this is to be expected with LLMs. But AI could encompass so much more with suitable input. But training it on physics and maths say is much harder and less eye catching. More likely a slow burn.

[jqpabc123]

But training it on physics and maths say is much harder and less eye catching.

I think it would be much more eye catching.

It would involve a machine learning to reason --- something LLMs just don't do.

[JSDevOps]

Yes, https://www.zerohedge.com/technology/openai-google-anthropic...

[rvz]

Yes it is hitting a wall. This article is complete cope.

AI model progress is hitting diminishing returns, even as the large AI companies have said and as predicted by Yann LeCun.

I guess the AI LLM grift is about to run out of road and headed into a new AI winter.