[Insanity]

That's what they'll sell you. But fundamentally current LLM implementations seem to be the wrong methodology for 'AGI' and I don't think we're remotely close to having this.

[dannersy]

Good luck convincing HN of this point. Half the replies in this thread reek of insanity.

[tptacek]

Don't sneer at imaginary HN commenters. It's always possible to a point more persuasively than that.

[Yopolo]

This stupid LLM is already smarter than a handful of people I know.

I don't think we have pushed LLMs to their limits already, its still progressing its still crazy good and its able to run even longer today than yesterday.

But researchers are already working on more architectures.