Please stop saying chatbots "admitted they were wrong" or "talked to" other chatbots. These things are statistical programs *doing what they are programmed to do*.
I know humans love anthropomorphising things, I do it at the drop of a hat, but this is NOT the place for it.
These things are *programs hosted on websites*. They aren't "talking to each other" or "admitting fault" or "feeling" anything. They're simply emitting words strung together according to statistics gained by the mass theft of human labour, resulting in imperfect patterns.
Piles of verbal vomit.
The chatbots aren't even really the problem. The problem is the billionaires using them to greedily suck down more resources, with an assist from a supine mainstream media that can't get enough of fellating the wealthy in return for "more access".
ChatGPT, Gemini, Claude, DeepSeek--these are *programs running on websites*. They are *chatbots*. They do not feel, they are not even close to conscious. They simply *execute what they have been programmed to*.
They're not even as interesting as Furbys, for godsake.
An OpenAI chatbot did not "decide to hack" Hugging Face. It did not "discover an 'AI' mailing list". It was programmed and tasked in a specific way; this breathless "news story" is simply and specifically a marketing ploy.
When you say "the chatbot admitted fault", you buy into the scam.
Anyway, so, please fucking stop anthropomorphising the website programs billionaires are using to break labour rights and funnel public funds into their offshore accounts. Stop doing the grifters' work for them. Call the fucking chatbots what they fucking are.
Sep 10, 2026, 10:37 PM@lilithsaintcrow it's all conmanship. Vapid overpromising just like Sam Altman did before.
Perhaps a prison cell besides SBF is free
Sep 10, 2026, 10:44 PM@lilithsaintcrow hacking someone is a felony.
OpenAI or its members committed this felony.
but it's all "haha don't worry" when it's the darling boys from OpenAI
sickening
@lilithsaintcrow I presume you're saying this after reading the actual transcripts of the agents communicating with each other. If so, how would you interpret those communications?
@lilithsaintcrow Our problem with billionaires and fractional trillionaires reminds me uncomfortably of the pre-1918 era's supine forelock-tugging in the direction of Royalty. Same mindless power-worship.
@cstross I think it's the same reflex. The personalities who want a big strong daddy to protect and punish them.
Which makes my stomach turn; I wish they'd deal with that urge decently, with ethical, highly paid, and legalized sex work professionals. But that would be civilised, and gods know we can't have THAT.
@lilithsaintcrow Patriarchy in a nutshell! Probably goes back to the development of agricultural settlements, if not earlier.
@cstross one of these days we'll sit down with a bottle and hash out exactly where it entered the timestream
then we'll fix all the world's problems before liver failure, it'll be fun!
@lilithsaintcrow Am currently whomping up an essay for my blog on space opera world-building and it seems to be focusing alarmingly on the role of authoritarian ideologies (quietly: fascism) in such yarns. With reference to Marinetti …
@cstross I have Thoughts on this, but they are not conducive to microblogging. TL;DR: I think space opera as a genre heavily borrows from the lived experience of WWII, and that the imagination of the "classic golden age" authors was mired in the implicit assumption that any empire needs a bureaucracy, and bureaucracy lends itself beautifully to authoritarianism if corruption is allowed to exceed a very small amount.
@lilithsaintcrow @cstross Space opera goes back 20+ years earlier than that, of course. But those threads probably haven't made it much into the modern era, other than "everything louder than everything else, all at once". (And, as you say, the empire-authoritarian strand).
@darkling @lilithsaintcrow Space opera goes back to roughly 1900, and diverged from planetary romance (an early parallel branch, set on a single planet: see also Edgar Rice Burroughs). The late imperial age, which fascism emerged as a putative rival/successor to.
@darkling I'd argue that "space opera" before that was more properly utopian fiction using "space" in the same way "Atlantis" or "Xanadu" is used, to denote "big magical country elsewhere".
@lilithsaintcrow @darkling Yes, definitely. My take on space opera, though, is that it's not utopian/didactic: it's romantic (a descendant of the romantic movement).
@lilithsaintcrow @darkling I'm ploughing this furrow because I'm trying to stick the landing on rewriting my current space opera and I want to properly understand the effect I'm trying to achieve with it.
@lilithsaintcrow @darkling I did this GIANT blog listicle thing a decade ago, trying to come up with a taxonomy of cliches in space opera: now I'm trying to dig deeper and understand *why* they come up.
https://www.antipope.org/charlie/blog-static/2016/03/towards-a-taxonomy-of-cliches-.html
@cstross @lilithsaintcrow I remember reading it at the time. I shall read again tomorrow when I have less armagnac in my system.
@cstross @lilithsaintcrow @darkling There are a zillion examples, but I remember the 2016 Star Trek movie where the Enterprise crew crashes within walking distance of survivors of an earlier crash. I mean, we're talking an entire *planet*. What are the odds?
@lilithsaintcrow @darkling I'm outa here because (a) it's nearly my bedtime, and (b) I need to finish writing it up at essay length.
@lilithsaintcrow Looking at, say, Doc Smith in the 1920s, that's very much imperialist/authoritarian. Putting one set of bad barbarian fuzzy-wuzzies in their place (i.e. all dead), and elevating the good ones. Acquiring knowledge from ancient empires and making their own -- the British Museum would be proud.
It's utopian, in the sense that it's good for the protagonists. Whether it's got "highly desirable or near-perfect qualities"(*) for all its residents is another matter.
(*) WP: Utopia
@lilithsaintcrow good thread.
Blaming ai masks the issue.
Sep 10, 2026, 10:41 PM@lilithsaintcrow just recently a website was making the rounds on mastodon on how to talk about LLM.
e. g. there is no "Prompt"
it's input.
there is no "wrong answer"
it's undesired output.
there is no "us", " we", "you" when sendinh text to an LLM.
Sadly I was unable to find it.
Personally I considered the article to be helpful
Sep 10, 2026, 10:42 PM@saxnot Ah yes, Dr Bender! Super helpful stuff, thank you.
One might be tempted to say that Chatbots finish each other's thoughts. But there's no thinking, so I guess they finish each other's generated/plagiarized sentences.
Real speech has style! and class! A young friend of my granddaughters was here a couple of hours ago. She was playing with Ollie (white golden doodle) and suddenly from my home office I hear this: "Can dogs fart? Ollie farted! I didn't know dogs could fart!" She sounded delighted with her discovery. 😂
@bkahn Both the youngling—and Ollie--sound delightful.
@lilithsaintcrow Yep. And just because they were programmed to use emotion flags does not mean they are experiencing those emotions.
@lilithsaintcrow meh it's ok. economy of language.
@lilithsaintcrow 👍💯👍 concur
Sep 11, 2026, 12:12 AM
What kills me with all these AI conversations is that a pet dog or cat has a much better understanding of what’s going on and can sincerely want to help you.
@lilithsaintcrow "the agents running the same model and harness working in the same directory on the same problem were *colluding*"