News for hackers. Stuff that matters.
Re:When AI Builds Itself: Our progress toward recursive self... by marcosdumay on Thursday June 04, 2026 @05:58PM (#48405173)
You need the AI eventually building another AI for the name to apply. This page is just bullshit. They vibe-code their harnesses, and yes, it shows.

Anyway, what does recursive self-improvement even means for neural-network based AIs? It's not clear it's possible at all.

Re:When AI Builds Itself: Our progress toward recursive self... by robbrown451 on Thursday June 04, 2026 @07:51PM (#48406245)
Where do you see evidence of vibe coding the harness? (aside from the name kind of implies it)

It seems odd to complain about a AI coding tool being coded with AI. That's just eating your own dog food. In my opinion it makes it better, because the tool is very well tested.

I'd also suggest that there are some things.... for instance the 3d piano thing.... that I doubt many people could do to that quality, with or without vibe coding it. I'd challenge anyone to do a better looking set of piano keys: vibe coding, old-school coding, or even Blender. Did you watch the video?

If you are against AI coding in general, you probably aren't in the market. Am I missing something?

In that sense this is no different than something like OpenClaw, right?

Re:When AI Builds Itself: Our progress toward recursive self... by marcosdumay on Friday June 05, 2026 @10:37AM (#48413152)
> and who are you talking about, Anthropic or the link I shared?

About Anthropic.

Re:When AI Builds Itself: Our progress toward recursive self... by ashdksnndck on Friday June 05, 2026 @12:17AM (#48407855)
Recursive self-improvement would be the model helping with the model research program. Coming up with hypotheses for training and architecture improvements, running experiments, interpreting the results, figuring out how to incorporate the best stuff into the next version, etc.