#BruceSterling

#BruceSterling

@bruces.bsky.social

Author, journalist, editor and critic combining his time in Turin, Austin, Ibiza, Belgrade etc #BruceSterling

9.5K followers2.8K following9.2K posts

*How come all these dead guys are dead? They all had standards of some kind, they didn't grant permission for that

Coleman Ridge@colemanridge.bsky.social

I am having trouble with the notion of standards that aren't human standards. What sets them and on what basis?

*Well, if he's Victor Frankenstein and the Emperor Napoleon, it's naive to assume that you would be allowed to merely subscribe and pay him off *Best case scenario: you're the brother-in-law *Eventually your son by the sister is the mayor of Tallahassee, Florida www.flickr.com/photos/bruce...

flickr.comJoachim MuratSoldier, French revolutionary, Marshal of Napoleon, King of the Two Sicilies, etc
Jim Infantino@jiminfantino.bsky.social

What’s the maximum you would pay in a subscription for access to a model?

*That's true, that troubled and kinky L'Interdit girl is indeed performed by Rooney Mara. But let "Lisbeth Salander, the brilliant but troubled computer hacker," land herself a lucrative perfume account, okay

Julian Sanchez@normative.bsky.social

Pretty sure that’s the actress Rooney Mara.

"I'm searching for a long-term, pleasant, joyously supportive marriage where the Forbidden Lover and I eventually congratulate one another about our many great-grandchildren"

This post is unavailable.

*Three, four years ago I would have been laughingly skeptical "Sure, cheap AI slop uber alles, but it's gonna have seven fingers and it can't count to five" "A pity it's not any better" *It got lots better. Not by human standards, but better by standards that are very hard to describe

*They're demanding $2T, will they get it, I dunno *Every industrial enterprise that ISN'T slop is visibly moribund, the oligarchy has leeched it all to near-death

*Well, if the "abundance of AI rapidly transforms every sector of the economy," then that "cauldron of boiling frogs" is rapidly transformed into a "Kettle of Planetary Slop" *Slop coast to coast, north to south. Every time-zone, every scale. Every demographic. Slop from a spoon-to-a-city

Paul McGhee@pmcghee.bsky.social

"Global Growth" ..... what even is that in a world that's a cauldron full of boiling frogs, owned and managed by Nazis?

*Planetary mayhem arrives at the breakfast table *Imagine being a "conservative" and trying to gaslight your "tradwife" about these stark kitchen realities *The contrails did it. Lesbians, Jews, somebody made the milk taste different

Paul Behrens@profpaulbehrens.bsky.social

We talked about climate's impact on food this morning on BBC One's Sunday Morning Live: imported food that has got more expensive (chocolate, coffee, olive oil), food hit at home (milk, onions & potatoes, lamb & beef), and wine changing flavour. And what we can do about it. Full interview below

*Whatever happens to musicians will happen to laid-off games journalists *Why is a society so saturated by computation so bored about its loud and shiny computer-games, there has to be some deeper cultural significance in this *Every games journalist should pivot to culture-critic

Aftermath@aftermath.site

When losing a job means losing a career: what comes next for laid off games journalists? aftermath.site/aftermath-game...

*Even if they end up as the Super Intelligent Super Conducting Super Collider, in some abandoned tangle of dewberry vines in Waxahatchie, that is hard to beat for gutsy ambition *Very Manifest Destiny. A robber baron's railroad, a Colt revolver and a Bowie knife, spit on the dice and roll 'em

*This must be the boldest public statement that I've ever read in my life. Your scheme is bigger than the Industrial Revolution plus the Internet, and it's happening right now, and you need 2 trillion!? *It's like Victor Frankenstein crowns himself as Napoleon

*Well, I can imagine an AI chess-engine that is *tuned* for fun *There are a lot of AI engines that are tuned to mimic the "styles" of human chess players *You would think there would be a market for a chess machine that doesn't remorselessly win, but always plays in the most-playful way

Gordon Hundley@drgoon.bsky.social

Martin Amis said "like a wall coming at you" and I think that's good. It's a level of play you can't comprehend. Every response is the strongest response, every risk you take opens a likely stalemate or defeat. I can't imagine it's fun.

*It will "hallucinate" and make that up, of course *Hallucination being the "crazy magical thinking ability" that is baked into LLMs themselves, from their get-go

John Smith@johnsmithrlperson.bsky.social

people assign all sorts of crazy magical thinking abilities to LLMs, one I run into a lot is people asking for super specific detailed knowledge of some recently released video game, or an older one that has been undergoing continuous development. like how do you think it can answer any of that

*Okay, but what is it like to *describe* a branch selection that is against board states *Because the human chess critics on YouTube, analysing ongoing AI games, never say anything remotely like that

Gordon Hundley@drgoon.bsky.social

AI is an inherently sloppy term. While modern chess algorithms include transformer models, the 'best' computer opponents use neural nets that represent all the states of the board. The branch selection is against board states, so your perception is apposite.

*You'd better hope that Karl himself doesn't go to Milan, and bring you back that particular gift because he has that in mind for you and your romance *If that's the case, red flags should shoot off in all directions and you should retreat into a romantasy novel

*No, you don't logically think-it-through in that Teutonic fashion, it's an instant, grab-the-product act of solidarity "Yes, this is the inner me here, yes, I'm forbidden and also brazenly-available while also being chained up in huge bleeding-heart rubies" *That awareness is worth the 80 euros

*You're Gretchen-from-Bremen, you're in Milan buying lingerie to take home *You contemplate the Interdit bottle *Do you engineer it, do you think: "I need to douse myself in this 'elixir' so I can be more 'forbidden' and play hard-to-get with Karl, my on-again, off-again, situationship guy"

*Also her new brand-extension is an "elixir" now, like a youth-extension elixir for the fading libidos of an aging demographic *And it's *provocative,* that is my point here. The dystopian element makes it impulse-purchase sexy

*A few meters away, the always uptight, wary-and-suffering "L'Interdit" perfume girl has been reduced to jewelled bondage *Not just forbidden, as she customarily is, but chained-up for it

*I don't think it's "ironic" that Demi Moore plays a horribly suffering beauty-victim in the movie "The Substance" and then retails a substance off the shelves at Sephora *On the contrary, this deliberately plays to a self-destructive element of masochism in the beauty racket as an industry

*It's become very common for humans to play chess along with an AI slider-bar estimating who is "ahead" *How would the AI itself become an AI chess critic, and verbally EXPLAIN to humans how it "knows" who is "ahead"

*An AI visual slop-engine will fulfill its prompt, but parts of the image that aren't prioritized will often decay into chaotic rubbish *But an AI chess engine always seems entirely engaged with every piece on the entire board. It's not choosy *Is that so, or does it just look that way to humans

*There are also a lot of fairy-chess variants where they're assigned absurd situations, like 15 knights vs 15 bishops, and they go about the process with their customary solemnity *Clearly chess-fans do enjoy these spectacles, but what's the nature of the appeal, exactly *Heath-Robinsonesque

*I also watch those chess videos, and they have many of the basic problems that art-critics have with AI artworks in general *The games are strangely dispassionate and drama-free. Nobody blunders. Nobody sets traps. *They strip away everything to win with the certainty of a one-pawn advantage

ponder@ponder.ooo

watching chess experts commentate ai chess games is funny bc they'll be rattling off all these ideas and consequences of why some move is clever or some other option isn't viable &c and then every once in a while they're just like "i dont rly get what this move is doing" and then just move on

*I blame science fiction dystopias for Demi Moore starring as a blood-gushing body-horror monster, while still managing to sell anti-aging cosmeceuticals #TheSubstance

*Hold on, Vivian Wilson is Elon's trans child and also that's a dead Tesla robot *Don't all rush to the Desigual all at once

#BruceSterling@bruces.bsky.social

*Tough break for that robot in the Barcelona airport. #Desigual *Also ominous that it's become genuinely fashionable to want to destroy them

*Well, I'm back in Turin. And here in my own neighbourhood, this brazen, shocking violation-of-the-norm *Is this a pre-revolutionary situation?! The very roots-of-society are stricken

*It seems a little late in the day to assume that they'll just peaceably receive their giant $2T crown of Saudi wealth-funds *Given contemporary geopolitics, I'd expect a situation more like relentless cyberwar zero-day attacks with the full-force of AI-assistance, plus drone-bombed data centers

AJC@acjc.mu.social

True, but open source/weight models are very capable too. I don't think OpenAI and Anthropic will have the duopoly they desire.

*Their "fulcrums of thinking levers" are not hard to see, okay? They're not metaphors, they're literally the size of small towns #DataCenters

Todd R. Rossman@mystic-scientist.bsky.social

If we shed anthropomorphic views of thinking, we can conjecture that AI will do something differently with concepts & symbols - how they are made, understood, & used. Symbols are our tools of abstraction to make the heavy physical work easier. On what fulcrum will they plant their thinking levers?

"We'll build the AI and then the AI will tell us how to make it profitable" *I get the logic in this, but what happens when the AI makes everybody else in the world *un-profitable* *Planet of Math Professors

*What does "formal verification of agriculture" look like *Once you've got those $2T Mars-sized machines set up, they've got to endlessly learn stuff to pay for their juice and their maintenance

dorian@doriantaylor.com

my mental model of formal verification is that it's orthogonal to anything "ai" does but then also imo ai code generation has a stupid amount of degrees of freedom

*Clearly it's beyond mere any mere "revolution" or "conquering empire" and is and better understood as some real-world singularity *But how do the lights stay on? How does it work? *Is everybody governed by "Anthropic's Constitution"? Is there a flag to salute

*What happens if they get away with it, in this "post-hyperscalar" world where AIs are bigger than national GNPs *Do they conquer all other forms of human endeavour the way they've just conquered mathematics *I guess that's on the agenda, but how does present-day society ever move into that state

*It's not like they never WARNED everybody, repeatedly, that it was getting more and more dangerous *Also, they've got serious uses for the 2 trillion, like, constructing giant humming databanks all over the landscape that are the size and shape of Mars colonies

*The problem there is not finding the boldness to shake down the whole planet, it's gonna be the rival protection-racket gangs *I should have figured this out quite a while ago. I feel naive now *I enjoy the sense of naivete, actually, it's rejuvenating

*Improving generative programming until it can detect every tiny human error in decades of human labor *And then holding the entire software industry hostage to "protect" them from threats that you created *Of course that racket is "worth 2 trillion," if you can get away with it

#BruceSterling@bruces.bsky.social

*It's as if the Grammar Police had invented a Newspeak with an accurate, fully-provable grammar, and if everybody in the world didn't promptly learn to write that way, they would all be attacked and robbed by the new Grammar Criminals www.anthropic.com/research/for...

*I blame science fiction dystopias

Faine Greenwood@faineg.com

it’s amazing how these billionaire tech freaks *love* mainstream sci-fi and superhero movies, and also manage to completely miss the incredibly unsubtle and heavy-handed “we need to protect human creativity and freedom” moral that essentially all those movies are imbued with

*It's rather dreadful to single them out by name before attempting to assassinate them #AI #datacenters #bombing #Chervonenkis #Galushkin #Lyapunov

*It's as if the Grammar Police had invented a Newspeak with an accurate, fully-provable grammar, and if everybody in the world didn't promptly learn to write that way, they would all be attacked and robbed by the new Grammar Criminals www.anthropic.com/research/for...

anthropic.comLLMs are making formal software verification feasible at a much larger scaleClaude formally proved security properties in the compiled code of 123 open-source libraries. Along the way, it found bugs that had eluded detection for years.