weekly reminder; happy Monday Have you considered NOT using AI?

yes, and this is also an especially Global North/Western perspective to people's hearts: their pets — which they care about much MUCH more than they care about other humans who happen to e.g. not be USAmerican or white

I tell you what, Ring figured out exactly how to weaponize white people’s empathy in service to a corporately controlled surveillance state. Tell them it will find your dog.

' Ronald Barnett describes the contemporary fascination with skills as a form of ‘bewitchment’ (others have put it more strongly elsewhere)....When individuals are defined in this way, they become more easily interchangeable in the labour market, increasing workers’ vulnerability.'

e/acc/divorced…

"[Learning] is not oriented towards immediate use and immediate disposal. Learning feels slow because it is digging a deep foundation[. I]t can withstand challenge, it inspires confidence." — @caitrionaleahy.bsky.social, Norah Campbell, @clarekelly.bsky.social www.irishexaminer.com/opinion/comm...

Thanks for sharing @olivia.science! In case the article is paywalled for some - here's another way to access: archive.is/TmkDk 😊

A conversation I often have with my MSc students is about the limitations of models. If science is the process gaining new knowledge about the universe. Then what knowledge have you learnt from a high performing model with uninterrptable parameters?

and because they cannot they will burn the planet instead, which yes, they were doing anyway, which also was not a coincidence bsky.app/profile/iris...

It's just statistics repackaged from long ago and it's not useful in this form except for furthering fascism

Classic bsky.app/profile/oliv...

Maybe the premise of a lot of "AI" is "What if we could get everything we wanted without having to do or understand anything?"

I have thee fortune to work with a small number of bright young people at school and university, and they will all hear my severe cautioning about use of AI. At worst, over-reliance will simply leave a hole in their brain where understanding should reside.

Have you read the academic paper by @olivia.science and others likening it to the smoking industry? It's an extremely good paper likening it to the smoking industry: in the same way they got the backing of doctors to cover up its harms, AI is 'capturing' parts of academia by offering bribes.

Thanks Olivia. I think the sign of a good analogy is that its strength means it sticks in the memory straight away. Both of these harmful industries (smoking & AI) are backed by multinational businesses with huge amounts of available capital and other resources deployed to get their targets hooked.

this is now published here doi.org/10.1037/rev0... (pdf: repository.ubn.ru.nl/bitstream/ha... ) but note especially rows "Question" and "Goal" in table 1 below bsky.app/profile/oliv...

"prediction is a red herring" is so hard for so many to grasp, especially those trapped in correlationist thinking bsky.app/profile/oliv...

sorry but openai bros ain't no maria skłodowska-curie

also radium is something that occurs naturally and was discovered, ai models are created on purpose, so yeah, not a great frame

Called artificial where content is faked, it often is merely assisted intelligence. For either, a more sophisticated level of literacy is urgently needed for information users.

The tech companies pushing AI on us are essentially drug-pushers. Same techniques. Same motivations. Same results: a bunch of mindless addicts unable to think for themselves and without a chance to unhook themselves.

This a million times. I would sooner eat glass than prompt an ecocide-plagiarism-harassment-psychosis radioactive waste machine. And that's how I've felt since the beginning of time.

As the Internet slowly drowns in AI slush, I’m finding that physical items are the best for all kinds of reference. From ‘Art of’ books and collectables to photographing things around me that peak my interest. Physical forms of reference are completely under our control. No nonsense.

#telekitneticart #formline #indigenousart 🚮⚙️ You cannot automate the human experience (Sticker up now) 👉 ko-fi.com/s/3ecf690ed7

Trust me, the woman of academia are not at all surprised by the number of academic men orbiting Epstein.

The irony is that they want to force user facing AI-agents on everyone, which is an unsafe nightmare, so they can brand it "AI", when they could just use "AI" under the hood to choose between safe, deterministic outcomes. I hate "AI", and even I see how much potential is wasted.

A huge hint here, can’t quite put my finger on it…

Effectively accelerated years of photos into the ether, never to be seen again.

They think if they pour enough juice in the plagiarism machine it gonna start giving efficient solution for the energy crisis and be able to continue the "climb".

There’s no way in hell that AI will lead us to solve those energy crisis issues, it’s just leading us further into the crisis

Also, I have yet to find a genuine useful application of GenAI in my life at all.

I don't speak for "we" but there's certainly lots of pre-boom examples. The thing that is new seems to be scale and that comes with significant downsides that supersede usefulness.

LLM AI doesn't have genuinely useful applications, genuine being the operative word.

I was going to say the difference was that people didn't know the harm radium did, but it turns out business owners did indeed know radium was harmful to their employees but covered it up until a lawsuit exposed them. en.wikipedia.org/wiki/Radium#...

I don't understand why so many self-proclaimed genAI "thought leaders" happily post their self-owns for the world to see and mock. Literally so stupid they can't even see that they're both (a) the problem in the scenarios they post, and (b) amongst the dumbest motherfuckers alive. 🤷‍♂️

But also not because Asbestos was used for centuries before the harm it caused was understood/campaigned against as it was in use for for hundreds years. this shit I've been saying saying is harmful and bad since back when Midjourney was only capable of pixelly smudges.

Oh and literally fascists in the example bsky.app/profile/davi...

Not the best frame since it's not new, and yes, it's a terrible idea to put it anywhere www.project-syndicate.org/commentary/a... bsky.app/profile/oliv...

e/acc, so Nazi too

e/acc/OSS 🙊

See also: * Mercury * Lead * Asbestos * Talc * CFCs * Arsenic * Copper * Cadmium

And web.mit.edu/slava/space/... doi.org/10.48550/arX...

It's just statistics repackaged from long ago and it's not useful in this form except for furthering fascism

"The purpose of computation is insight, not numbers" -- Richard Hamming

A reminder that Kate Manne and I have a lecture coming up at Princeton…

NYT LLM MLM BS

To be very clear, that lady in the NYT article is not making any money with her hundreds of AI slop books. She makes money by *convincing people* she makes money that way and getting them to pay her to teach them how to do it. It's an evolved MLM, and the NYT is helping market it.

This gets at something important: what is a model for? Is it to tell you the future, and you believe it and act accordingly? Or is it to help you understand how things work so you understand what you're doing?

but the main point is that I think you need to understand a problem in order to solve it effectively code generation tools just generate likely code based on other problems people solved. if you have a common problem your odds of usable code are decent, but that's not because it understood anything