What kills me is that there are so many obvious ways to be less wasteful about this. Nondestructive book scanners exist. They’re expensive but it’s not like AI companies are averse to throwing money down a fucking hole. Even if that’s not an option, it’s possible to rebind the pages and return the books to the market. And regardless of what happens to the physical books, the scans they create could be archived in a format that could be available as a digital library rather than just being fed into the statistical meat grinder to keep the trough topped off with slop. Even if they’ve got to fight with copyright holders it would cost them basically nothing to leave the option open and given the PR battle they’ve been losing it feels like doing so would be incredibly obvious. Hell, even Google Books was able to navigate this in a way that was less cartoonishly evil than this because they could point to their digitization effort as a public good in ways that Anthropic here just fucking can’t.

how productive the field will become

Let’s just ship all these people to Macroeconomica

I mean, I think the most charitable interpretation of the goal is effectively the one from Hitchhikers Guide to the Galaxy, where all the AI systems including the simple automatic doors are fully sentient and have simply been given the ability to feel either euphoric joy or simple boredom, and then they have a chip that activates happy mode when they fulfill whatever purpose they’ve been given (“It was my pleasure to open for you, and it is with deep satisfaction that I close again behind you.”) At that point, the argument goes, not making them do the thing would be cruelly consigning them to eternal boredom. This is also the same series that proposed solving the ethics of meat consumption by genetically engineering a cow that actively wants to be killed and eaten by restaurant patrons more than anything else.

If anything that makes for an even stronger case. Gangstalking is already a known pattern of delusional behavior and if a service pitched in part for its ability to provide relief is actually reinforcing that…

I don’t think anyone here is likely to disagree that current LLM systems can’t coherently be said to have some kind of internal experience and the design and functioning of those systems seems to fundamentally preclude the kinds of conscious experience that we want to assign moral agency to. But the key to understanding this as a moral issue imo is that the people designing and building these things seem like they really want to create something that ought to merit that kind of moral consideration. For all that the humanizing language of agency, intelligence, welfare, etc. is in reality part of the grift, I think there’s a strong catch-22 to be argued that if it’s not part of the grift then slavery is pretty implicit in the value proposition of these companies.

The actual worst man, and also just fucking sad. Imagine knowing that not only did you peak in high school, but that you peaked specifically in the moment when you fucked up and lost the state championship. Like, you can just smell the empty beer cans and tears piled up around this guy’s posts.

This was a fun one. It feels like there’s a shared affinity between Rats and traditional conservatives for confusing length and volume of output with actual depth or quantity of thoughts, and I deeply appreciate your willingness to wade through it all. Now that I think about it it would be difficult to write 500 pages saying basically nothing and not repeatedly blatantly contradict myself.

Also, release the Foucault tangent, you cowards!

Ooh. I love these. I’m “I don’t think Sam Altman or most of his social circle are good people and I don’t want to give them more wealth and power.”

Vibe Construction

2027 was the original pitch for an imminent posthuman apocalypse. 2040 is an attempt to argue for policies that will slow down enough to avert said apocalypse. It’s all the dumb of AI 2027 but with some shockingly bad geopolitical analysis and promises that if we do what they say we can have what I can only describe as a neoliberal manifest destiny space program.

I feel like there’s a version of this that has a slider for how aggressively you’re willing to sacrifice readability, and you could probably get pretty decent results on the scale of 2x to 2.5x just using different encodings of the same basic glyph.

I know we’re all familiar with the metaphor of LLMs as roulette wheel, but I think there’s some real craft to turning that into Russian roulette.

I have definitely done this. Written half an essay in the awful.systems comment box, realized I don’t know what the fuck I’m saying anymore, and deleted it. I like to tell myself that forcing myself to realize I was being incoherent is enough to justify the costs. I guess if I acknowledge that I don’t rely on my online footprint to put food on the table the way the proper blogonauts (we’re still doing blogs, right?) do I can understand the incentives against doing that though.

I think this particular pattern is another consequence of long-termism in the water supply. It’s easy to disconnect yourself from humanity as a concept when you’re in a space that often assumes that you and everyone you could ever have a personal relationship with are already long dead. The fact that this makes it easy to ignore the problems that people today are having in front of your face is purely coincidental.

It’s been a long time since I’ve seen basic questions about functionality answered by “something something blockchain” but that appears to be the answer here.

You know, I think there’s an argument to be made for death threats against tech oligarchs being a crude form of direct action towards redistribution.

I can’t decide if it’s fitting or ironic that this group of men who spends collectively one week a year in spaces they don’t either literally own or otherwise totally control has such a hard time understanding concerns the rest of us have about privacy.

I should probably be grateful that this sent me down the road of learning about pet trusts, the actual legal vehicle for leaving large sums of money to an animal. But that’s less because it’s relevant to the point here and more because I’m a fucking nerd. Notably the money is controlled by a trustee who is empowered to pay for the animal’s food, healthcare, etc. I imagine that an enterprising trustee in the most famous cases could make some very compelling arguments about the necessity of maintaining, say, a big house for the cat to roam around in that they and their family would coincidentally live in as well, but I don’t care enough to look into it in-depth. If the animal dies before the trust runs out of money, then the remainder is distributed according to either the trust instructions or applicable state laws (in the US, at least). Of course the cases of people putting millions into pet trusts are usually less about the pet trust and more about spiting the other presumptive heirs, a sort of disinterestance with a more spiteful edge.

To the actual argument here, however, all I can say is that nobody without a large estate has ever made the allegedly absurd decision to dedicate their estate to the care and keeping of their pets. Every single case where that has happened has been because of a member of the supposedly-enlightened moneyed classes being fucking weird and having too much money.

I remember having a lot of fun with Roborosewater where they could generate something that matched the format of a Magic: the Gathering card perfectly and sometimes even something playable within the rules but it was always nonsensical in some ridiculous way. There’s a couple streams from the LoadingReadyRun gang drafting a cube from the resulting chaos that are a hoot.

Ed: Here’s a sampling in video form because I can’t find a “best of” article and they get into the “analysis” a bit.

If you look at literally page 1 you’ll find the first footnote pointing at another chapter of the same goddamn book and now I’m going to go weep tears of impotent rage.

Ed: sorry, it’s page 8, which is the first page after the self-congratulatory preface.