OK I sped read that thing earlier today, and am now reading it proper.

The best answer — AI has “jagged intelligence” — lies in between hype and skepticism.

Here’s how they describe this term, about 2000 words in:

Researchers have come up with a buzzy term to describe this pattern of reasoning: “jagged intelligence." […] Picture it like this. If human intelligence looks like a cloud with softly rounded edges, artificial intelligence is like a spiky cloud with giant peaks and valleys right next to each other. In humans, a lot of problem-solving capabilities are highly correlated with each other, but AI can be great at one thing and ridiculously bad at another thing that (to us) doesn’t seem far apart.

So basically, this term is just pure hype, designed to play up the “intelligence” part of it, to suggest that “AI can be great”. The article just boils down to “use AI for the things that we think it’s good at, and don’t use it for the things we think it’s bad at!” As they say on the internet, completely unserious.

The big story is: AI companies now claim that their models are capable of genuine reasoning — the type of thinking you and I do when we want to solve a problem. And the big question is: Is that true?

Demonstrably no.

These models are yielding some very impressive results. They can solve tricky logic puzzles, ace math tests, and write flawless code on the first try.

Fuck right off.

Yet they also fail spectacularly on really easy problems. AI experts are torn over how to interpret this. Skeptics take it as evidence that “reasoning” models aren’t really reasoning at all.

Ah, yes, as we all know, the burden of proof lies on skeptics.

Believers insist that the models genuinely are doing some reasoning, and though it may not currently be as flexible as a human’s reasoning, it’s well on its way to getting there. So, who’s right?

Again, fuck off.

Moving on…

The skeptic’s case

vs

The believer’s case

A LW-level analysis shows that the article spends 650 words on the skeptic’s case and 889 on the believer’s case. BIAS!!! /s.

Anyway, here are the skeptics quoted:

• Shannon Vallor, “a philosopher of technology at the University of Edinburgh”
• Melanie Mitchell, “a professor at the Santa Fe Institute”

Great, now the believers:

• Ryan Greenblatt, “chief scientist at Redwood Research”
• Ajeya Cotra, “a senior analyst at Open Philanthropy”

You will never guess which two of these four are regular wrongers.

Note that the article only really has examples of the dumbass-nature of LLMs. All the smart things it reportedly does is anecdotal, i.e. the author just says shit like “AI can do solve some really complex problems!” Yet, it still has the gall to both-sides this and suggest we’ve boiled the oceans for something more than a simulated idiot.