From the preprint:

The key formula (39) for the amplitude in this region was first conjectured by GPT-5.2 Pro and then proved by a new internal OpenAI model.

“Methodology: trust us, bro”

Edit: Having now spent as much time reading the paper as I am willing to, it looks like the first so-called great advance was what you’d get from a Mathematica’s FullSimplify, souped up in a way that makes it unreliable. The second so-called great advance, going from the special cases in Eqs. (35)–(38) to conjecturing the general formula in Eq. (39), means conjecturing a formula that… well, the prefactor is the obvious guess, the number of binomials in the product is the obvious guess, and after staring at the subscripts I don’t see why the researchers would not have guessed Eq. (39) at least as an Ansatz.

All the claims about an “internal” model are unverifiable and tell us nothing about how much hand-holding the humans had to do. Writing them up in this manner is, in my opinion, unethical and a detriment to science. Frankly, anyone who works for an AI company and makes a claim about the amount of supervision they had to do should be assumed to be lying.

Someone claiming to be one of the authors showed up in the comments saying that they couldn’t have done it without GPT… which just makes me think “skill issue”, honestly.

Even a true-blue sporadic success can’t outweigh the pervasive deskilling, the overstressing of the peer review process, the generation of peer reviews that simply can’t be trusted, and the fact that misinformation about physics can now be pumped interactively to the public at scale.

“The bus to the physics conference runs so much better on leaded gasoline!” “We accelerated our material-testing protocol by 22% and reduced equipment costs. Yes, they are technically blood diamonds, if you want to get all sensitive about it…”

IEEE Spectrum publishes a column saying that Wikipedia needs to embrace AI to avoid the dreaded generation gap, gets roasted

https://mastodon.social/@ieeespectrum/116059551433682789

Pointlessly insulting, cruel, assumes total incompetence at life rather than a momentary mistake in managing the information overflow, juvenile in the bad sense of the word.

object level issue

<Kill Bill air raid sirens.mp4>

Previously posted in the stubstack, twice. See threads for earlier sneers. No objection to making a top-level post and gathering further sneers, of course.

The idea that a government from the actual McCarthy Era would be adept at handling an organized labor response to massive upheaval in the job market is… what’s the superlative of “lolz”?

I do believe that’s literally how the automation dystopia began in Vonnegut’s Player Piano.

“Can’t read” is the kind of insult we don’t need in this context.

https://xcancel.com/allTheYud/status/2020177765534953662#m

And when our descendants build a museum to these bizarre and painful times, there will be docents of them. Docents!

“There are dozens of us! Dozens (fractional)!”

A polycule with Aella, otherwise known as a nightmare fuck rotation

Awful.systems is not debate club. Nor is it peer-review club. No one is obligated to nitpick individual sentences in a preprint or erect monuments of text about details within it, particularly when a discussion of the broader failings of the “research” culture in that area is more interesting, valuable and on-brand.

How much do people actually “like to claim to have read” books, rather than saying they want to read more big books but never have the time?

There’s a letter in the book of Asimov’s correspondence that his brother edited where Asimov says that he’d been asked “How close are we to George Orwell’s 1984?” again and again in the years leading up to 1984, to the point that he was sick of it and dreading the actual year 1984, when no one would ask him about anything else. I figure he had a lot of venom built up in his system that came out here.

He was also a veteran of science-fiction fan club drama, after which he worked in academia, so yeah, he knew sectarian in-fighting.

Ryan Mac:

Epstein had many known connections to Silicon Valley CEOs, but less known was how he made money from those relationships.

We did a deep dive into how he got dealflow in Silicon Valley, giving him shots to invest in Coinbase, Palantir, SpaceX and other companies.

For example, here is Coinbase cofounder Fred Ehrsam in 2014 emailing w/ people around Epstein, including crypto entrepreneur Brock Pierce, asking to meet Epstein before the financier invested $3m in Coinbase.

Coinbase was a two year old startup. Epstein netted multimillion dollar returns from this.

Here is Epstein asking Peter Thiel if he should invest in Spotify or Palantir. Thiel was (and still is) Palantir’s chairman and tells Epstein there is “no need to rush.” This is one of several emails where Thiel gives Epstein advice.

Epstein later invested $40m into one of Thiel’s VC funds.

One of @ering.bsky.social’s great file finds: Epstein tried to help create an tech fund shortly before he was arrested in 2019 with two tech types. One of his partners, however, was worried about the “optics” of telling founders that Epstein was involved.

So they suggested Epstein conceal himself.

At the end of his life, Epstein had assets of around $600m. A large part of that was due to his ability to get in early to hot tech deals. The returns he made off those deals helped fund his lifestyle.

[…]

While reporting this, I had something happen that’s never happened. A comms rep for one of the co’s disputed my reporting and said what I was telling them was untrue because it was not in Grok, xAI’s chatbot.

I was looking directly at the files. And this person was using AI to challenge the truth.

https://bsky.app/profile/rmac.bsky.social/post/3me4wmrgic226

I think Dominic Housatonic developed into a mushroom cloud as it progressed (past the interval captured by the slowmo).

The atomic bombings of Hiroshima and Nagasaki made mushroom clouds.

In the Baker test at Bikini Atoll, the bomb was underwater, dimming the initial flash and making initial transient effects more visible.

I was going to say that I had looked up Scott Aaronson in the files, and my conclusion overall was that nothing in them actually made him look worse than anyone already sees him. Joscha Bach name-dropped him as an interesting person (so what, really). Aaronson and Seth Lloyd each met with somebody who was working for Epstein (Charles Harper), at which there was some talk of making a “Cryptology in Nature” conference happen. As far as I could tell, that conference never did happen. It wasn’t even evident from Harper’s e-mails that Epstein had even been named at or before the meeting. I don’t think Aaronson could be blamed for having a business lunch with somebody who had been a big wheel at a private foundation (Templeton, in Harper’s case) and who said he could get private-foundation funding for a meeting in Aaronson’s subject area.

And then Scott Aaronson had to go and write a blog post about his being in the Epstein files. Short version: He says he had lunch with Harper, after which Harper wrote him a follow-up that named Epstein “for the first time”, and then he ignored Harper after hearing about Epstein’s conviction. That sounds consistent with the “no real harm, no real foul” impression that I would have been willing to endorse after searching the e-mails myself. But then the epilogue! Scott comments on his own post:

I had a further thought. Back in 2019, when Epstein became a central topic of conversation following his arrest and then death, and lots of my scientific colleagues were telling stories about their contacts or near-contacts with him, it struck me that there were zero stories about any scientist—liberal or conservative, male or female, morally naive or morally astute—saying, “no, of course I want nothing to do with you, because you’re friggin’ Jeffrey Epstein, the infamous mass rapist!”

So I concluded that, if anyone now imagines that they would’ve responded that way, it’s almost certainly pure hindsight bias. Indeed, even after Epstein’s first conviction, a short jail stint in one’s past for “soliciting prostitution” simply doesn’t sound disqualifying, according to the secular liberal morality that most academics hold, unless you researched the details, which most didn’t.

Meanwhile, in 2019:

Penrose and Epstein had met at a June 2017 conference on the science of consciousness in San Diego. “Although the topic [of consciousness] is not what I do, when I saw the list of speakers and was offered a plenary talk, I decided that it would be a good thing for me and a good audience to hear about my experiment,” says [Ivette] Fuentes, a professor at the University of Nottingham in the United Kingdom whose work is supported by the Penrose Institute.

Shortly after returning home, Fuentes says, she and Penrose had a conversation. “Would I be interested in receiving funding from a wealthy man who had also been convicted of a sex offense?” Fuentes recalls Penrose asking her.

Fuentes immediately said no, citing ethical objections, and quickly forgot about the conversation. But 2 months ago, after reading that Epstein had been arrested, she called Penrose. “Was it Epstein?” she asked him. “And he said, ‘Yes, I think it was.’ And I said, ‘Oh God.'”

I dunno, Scott. Maybe you should find better friends.