Подтвердите e-mail

Для публикаций, комментариев, реакций и сообщений подтвердите адрес.

Публикация

I'm fascinated by this new form of PR that's emerged with LLM companies where they leak stories like 'our product spontaneously assembled an army of domestic appliances & announced it would exterminate the lowly hairless apes - we barely managed to pull the plug' to imply it's really powerful

Обсуждение

Ответы

Cal Newport recently wrote a great piece about this www.nytimes.com/2026/06/17/o...

Opinion | Dear A.I. Companies, the Doom Trolling Needs to Stop (Gift Article)The major A.I. companies keep telling us how dangerous their new models can be, yet they must keep on building them.www.nytimes.com

It's weird because anywhere else in software development we call those bugs. Apparently when rich special grifter boys do them they're special and good and the grifter boys should be given more of our pensions.

Once we have safely contained it, just you wait, like so much productivity, lots and lots of innovation, at super low low costs.

It was funny when OpenAI publicised ChatGPT hacking huggingface and then Anthropic immediately had to say "Ours did too!" and then OpenAI are like "It actually happened loads with us" It's like two kids talking loudly in front of the rest of the class but the rest of the class are shareholders.

Ответ для Niall O'Sullivan

AI is just "my uncle who works at Nintendo" for grown-ups.

Maybe this existed before, with defence contractors or something? But you'd never get Unilever leaking stories where Domestos burnt through six floors of a block of flats & melted a cat's face after a cleaner missed the toilet bowl with it

Ответ для Tim Clare

I might start circulating accounts of terrifying near-misses involving my books via press release. Wait... is this what the Necronomicon & The King in Yellow were doing all along? Did we get doom-trolled?!

Ответ для Tim Clare

Kellogg's demand government spend more on Pringle-Addict Rehabilitation Centres.

I’m trialling AI tools where I work. Here’s a US tech trade body report: “Hugging Face experienced a high-velocity, multipolar, autonomous AI cyber attack where photos of some staff, whose details it hacked, were published as non-consensual topless deepfakes on their social media” Spin that, baby‼️

Yees! Also I noticed (as below) completely uncritical verbatim quotes used as headlines. To quote Hbomberguy... This is journalism, yee haw! bsky.app/profile/holl...

hollyashtree

Love that newspapers uncritically begin headlines with "Anthropic says [insert nonsense here]" Literally any other brand would kill for that kind of PR. "Waitrose says using any other supermarket will make your guts spurt through your nose."

i thought the entire point of this LLM mass hysteria was to prevent the 'ai' from becoming evil and destroying the world.

Ответ для iucounu

Oh wait - I know where I've heard this marketing before. PUAs shilling 'dark psychology' courses with hypnosis/NLP/persuasion techniques.

bizarre to me how many "journalists" just uncritically repeat anything a tech CEO says. if the CEO of McDonald's claimed their burgers cured cancer nobody would report it with a straight face, but "we will build a Dyson sphere to power our chatbot" is somehow considered news

[record scratch] “—that’s right, it’s me, Roko’s basilisk. You’re probably wondering how I got in this crazy mess…”