LLMuddites

Had one of those glum moods when a promising Nathan Tankus piece stopped short due to incapacity to subscribe with “$”.

He never gets to the connections promised for us on the free tier. (Course, I’d subscribe if I had a decent wage. Nathan is one of the best MMT journos around.)

So I had to write my own piece so I could read something half entertaining and half serious 𝓐𝓕.

Robot Booboos

🤖

In related: I’ve read nominal lefties recently tout “Ai” as a human boon, claiming we are just not getting tech right because of “techno conservatism”. Sure, sure, bro. But nah.

I am not afraid of technology, I am afraid of what people can do to us “little people” when they hoard power. (Lilliputians/Jonathan Swift, all that classic metaphor hitting us in the face still, a century later).

I guess techno conservatives exist. But these are not luddites. Many so-called Luddites broke equipment mainly for protest, not because they feared technological change. It was the “boss” forcing them into dehumanizing work that was the issue, not “the machines”.

The pomo/modernist dumb-dumb liberal-normie use of “Luddite” as a term for technophobes misrepresents the original movement; the historical Luddites were skilled individuals who understood technology but resisted its use to disenfranchise workers.

So people, let’s not disenfranchise workers, then we can enjoy the fruits of labour saving technology. But also, let’s not eat up oodles of electricity and guzzle drinking water for running “Ai” servers so you all can type in ChatGPT and Sora prompts like fools.

When you are using the Ai servers just have a think about how many people you are harming first. People who are living near those server farms and who have been thoroughly disenfranchised, you will likely never hear their cries nor see their tears from your living room comfort.

Dectectivizing Tankus

I’ll have a go imagining what Nathan wrote below the free tier.

Firstly, while not using as few words, I think the Trump Adminstration is a robot, getting policy form LLM’s, but the Grok type, i.e., sucked up a whole lot of 4chan, 8chan and reddit in their LLM training data corpus.

Hey, nerd guys,… guys! A text corpus is corpulent for good linguistic etymological reasons. It’s dead stuff. It’s not going to Frankenstein into life and sentience.

I “get it” though. You wannabe scientist. So you believe ideologically in physical materialism. Well, sorry for you, because physicalism can never account for subjective phenomenal qualia. It is bad science to have a rigid ideology. The whole decrepit concept of “Emergence” is thorough nonsense. Get off reddit and start thinking for yourself.

Let me give you a hint about what I do — as a physicist.

I take physicalism seriously when I explore physics.

Like a marine biologist takes the ocean seriously. It does not 𝔘𝔠𝔎𝔦𝔫𝔊 mean life on land is nonexistent.

That’s all you need. Don’t swallow Lackoff & Johnson on mathematics, for heavens sake they are like.. bourgeois 𐝥ꗍꖡꗇ𐝥ꕒꗍꕒ on that topic. Then if non-physical realms exist (like, say, mathematics is platonic, which it obviously is to anyone with more than peanuts for brains) then

On the “Ai” stuff, there’s a whole lot of worms there that I try to sometimes write about, but I feel I am swimming against a normie tsunami of nonsense. People too easily psychologically project their consciousness onto a bunch of text in front of their face in real-time.

Hard to blame them. In all history not text before has ever been machine generated at inception (from the typographical primitives). So we’ve managed to build such technology. It should be cause for celebration, but instead the technofascists are out convincing otherwise nice people that humans are nothing but meat machines and will become a redundant species. It’s extreme fascism, a type not on Umberto Eco’s list.

Why are people so willing to dehumanize themselves?

I thought the bosses were good enough at that we did not need to add computer code output in to boot (several layers of small furry metaphors gathered together in a commandline and grooving with a pict there!)

Hey, reddit guys, there’s nothing wrong in admitting the corpus sum of all past Internet text deemed worthy can stochasticaly out-perform your brain cells on writing memos and sometime executable softwares. No shame at all. But it does not mean the machine the code is running on is conscious. So stop lauding the phantasmagorical future Superintelligent Ai overlords. Just stop it, because that is what is unbecoming an intelligent person.

Behaviourism

Having got that bit off my keyboard, the conventional “wisdom” is that machines can be conscious. That’s just stupid. Humans are not machines, and machines have no souls (there is an easy proof for this). But if you make judgments about “has a soul” based on input-output behavioural responses, you are no less 𝘤🅁🅴𝙩ⓘ𝕟ous than Burrhus F. Skinner. Does anyone remember that ꖀ𐝥ꗍꖡꕯꖦ? Seems memory is short among the techno-optimists.