Tzt

Researcher: hack me

Model: I committed a crime

Researcher: oh my god

This implementation has an explicit mutation rate! That's not in the spirit of the original paper, where programs "mutate" from interacting with other random pre seeded programs.

Greg Egan's "Rifters"

By Peter Watts actually.

"free will" also known as digits of pi mod 2

Yes, absolute majority of new ones use CoTs, long chain of reasoning you don't see.

Also some of them use such a weird style of talking in them e.g.

o3 talks about watchers and marinade, and cunning schemes https://www.antischeming.ai/snippets

gpt5 gets existential about seahorses https://x.com/blingdivinity/status/1998590768118731042

I remember one where gpt5 spontaneously wrote a poem about deception in its CoT and then resumed like nothing weird happened. But I can't find mentions of it now.

In that analogy "someone" is an AI, who of course switches from answering questions from humans, to answering questions from other AIs, because the demand is 10x.

I agree with this. This a remarkably bad podcast. And also pretty bad paper to focus on. As the podcast was quite bad, I just read it and it was about nothing at all.

Like, it's a basically blogpost that muses about uhhh couple examples it pulled at random from esolang wiki and has literally no point. Beside prescriptive one. Formatted as a paper, which I admit takes some skills.

Well, it's also an indicator for how well its other claims would hold up if you dug deeper on them too.

Well, there are also legless salamanders, that look like eels pretty much.

https://en.wikipedia.org/wiki/Two-toed_amphiuma

Some of them have no lungs even:

https://en.wikipedia.org/wiki/Microcaecilia_iwokramae

So, it makes sense to say that eels are fish, because there are lungless eel-like creatures that are actually amphibians.

I don't get it, does the prediction go backwards or forward along CA generations?

CNNs are CA if you don't insert fully connected layers, actually.

Well, kinda? I often know what chunks / functions I need, but too lazy to think how to implement them exactly, how they should works inside. Yeah, you need to have overall idea of what you are trying to make.

What do you mean several years, it became feasible like 6 months ago lol. No, gpt3.5 doesn't count, it's a completely useless thing.

"Wild mammals"

nice one, journalism

why not make AI on the same platform as the human brain? what is so exceptional about it, and even if it is an exceptional material, why not just use it?