Zitron's empirical attack on AGI

This is how a god dies, not with a cosmic explosion, not with a descent into eternity, but with the face of a pissed off Englishman calling it a big pile of bollocks.

If five years ago you had asked me for a single symbol to express the first meeting of Humanity and AI, I might have said HAL’s red eye from 2001, Keanu Reeves awakening from The Matrix, even the Terminator crushing human skulls.

I would not have suggested the face of Ed Zitron.

But here we are.

Ed’s expression captures the mix of “what the f*ck is this sh*t” and “I am incredibly bored now” that I think most people feel today when talking about LLMs.

The Face That Broke the Mythos

Now you may be wondering – why is Science Fiction with Damien Walter dedicating a video to the technology critic Ed Zitron?

And why am I sitting on a toilet?

The answer to that second question is because I am in production on a major new 2 hour video essay on Skibidi Toilet and the toilet is part of the set.

And I’ve taken a detour to talk about Ed Zitron because Ed has strayed into the mythos of scifi.

Listen to almost any analysis of AI in the four years since ChatGPT hurled LLMs into the public consciousness and you’ll have heard a journalist or pundit say “it’s like something out of scifi”.

And then just carry on, entirely failing to analyse what it MEANS when hundreds of thousands of people say something is “like scifi”? Which might be why these pundits are four years late noticing that “AI” is…bollocks.

It’s a standing joke now that our technofeudal overlords read but don’t understand science fiction. Jeff Bezos thinks O’Neill cylinder slums-in-space sound cool. Elon Musk the hypercapitalist can’t stop citing the uber socialist Culture of Iain M Banks.

Technocapitalism and the Plunder of Sci-Fi

But capitalists don’t need to understand scifi to plunder it.

As our capitalist economies have stumbled ever deeper into crisis after the collapse of 2008, business has stopped being about numbers and become about stories.

In the years of zero interest rates capital went to he who had the best story. Electric vehicles. Space rockets. Missions to Mars. Neural links. Elon Musk became the world’s richest man by being its most audacious teller of scifi tales.

As stock markets have soared and ordinary people have been immiserated, the scifi tales have grown taller. The Singularity is here, the light of human consciousness is expanding to the stars, the God of AGI is coming.

All stories born in scifi.

From HG Wells to William Gibson, Asimov, Le Guin, Octavia Butler and hundreds more, scifi writers have a created a new mythos for our age of science and technology.

A mythos that tech billionaires have looted for profit, turned into Silicon Valley bollocks, all now being called time on by Ed the Godkiller.

Yes. Ed has the numbers. But the numbers were always bollocks. Ed’s real weapon is Ed’s face. The face that says this story of summoning the AGI GOD isn’t just wrong. It’s embarrassing.

And an embarrassing myth is a dead myth.

Zitron almost single handedly broke the AI booster / doomer narrative cycle. LLMs aren’t going to save or destroy the world. In fact they aren’t going to do much of anything according to Ed.

His reward for changing the narrative is that the Zitron talking points are now being plagiarised by the talking heads on corporate media like the NYT.

Mainstream journalism is full of smart, educated people, who have also been brainwashed by neoliberalism since birth and are mesmerised by today’s scifi technobullshit.

Watching them stumble over the realisation that if AI is bullsh*t then, yes, the entire system they sheepdog for is bullsh*t, is at least amusing.

Collective Intelligence vs. The Machine God

As an avid Zitron fanboy reading between the lines I suspect Ed has a wider critique of the neoliberal corporate sh*tshow we’re all now forced to live in. The worst thing about capitalist realism is that it’s real.

But I don’t fully share the Ed Zitron take.

I suspect Ed’s financial argument will be proven 100% correct. But the slide from “the economics don’t work” to “the product is useless” is unnecessary. LLMs can be widely used and still entirely fail as the saviour of capital.

LLMs are a useful tool. They do a very large number of small, useful things that will likely be part of most people’s work. I used Gemini, for instance, to up-res Ed’s face from a screencap to a portrait. Useful. But on a scale of tech progress the magnitude of LLMs is somewhere between the synthesiser and the library.

Libraries were, for thousands of years, tools for the elite capture of knowledge. Libraries preserved knowledge, but only for those allowed within their walls. It took mass literacy and the Public Library to make the technology of the library emancipatory.

The technology is never the problem. Who owns and controls the technology is.

Hence why I describe LLMs not as AI, which summons fantasies of bad scifi movies, but as CI : Collective Intelligence. The intelligence is all ours, LLMs are just a new way to collectivise it. And a new way to collectivise our human intelligence is…useful.

Not the AGI machine GOD.

The promise of AI is not a useful new tool. It was needed to transform the economy. Because the economy is f*cked. It’s been openly f*cked since 2008. And fundamentally f*cked for much longer.

Global capitalism is a massive pile of debt built on promises of infinite growth that could never be kept. AI was the latest “hyperscale” scifi fantasy to keep investors locked into the delusion of endless growth.

Who Owns the Means of Production?

TRILLIONS of dollars have been invested into what is, in effect, a psychotic delusion of summoning an AGI GOD to solve all our material problems.

If Ed Zitron is right, and I believe he is, those trillions are gone. Converted into silicon wafers and stranded as fast depreciating assets. And the shock of that truth as it emerges might finally collapse the ponzi scheme of global capitalism.

But after the private equity firms and pension funds collapse, LLMs will still exist.

If you take the internet and stuff it into an LLM, it still does some somewhat useful things by giving us a new way to collectivise human knowledge.

The major problem with CI is who OWNS the knowledge of humanity. LLMs are being used by the sick, dying power structures of corporate capitalism to once again claim ownership over the human commons.

Our existing copyright laws have been broken ever since the internet arrived. If copyright law crushes Anthropic or OpenAI the ownership of CI will pass to Disney. Only the very naive can believe that will be a win for “creators”.

We’re deep in a multi-decade process of capitalism claiming the intellectual commons. Capitalism claimed the physical commons when Enclosure booted peasants off the common land and made us wage labour. The enclosure of the internet as web 2.0 was already a massive loss of the commons, LLMs are another.

Which is why I disagree with Zitron and other critics of LLMs themselves. Boycotting LLMs is walking off the battlefield. The corporations are 100% happy for you to not claim the Collective Intelligence of humanity. They will just take it all without a fight.

AI is not the problem. AI owned by capital is the problem.

“It took both time and experience before the workpeople learnt to distinguish between machinery and its employment by capital, and to direct their attacks, not against the material instruments of production, but against the mode in which they are used.”
Capital Vol. 1, Chapter 15, Karl Marx.

Marx here is describing how those newly “proletarianised”, made jobless, by machines always target the machines. Today it’s the knowledge workers newly proletarianised by AI making this mistake.

The machines, as Karl Marx understood, are not the right target if the workers want to win our emancipation. The challenge ahead for humanity is not fighting a Butlerian Jihad against the machines.

The real fight is the same as it ever was.

Who OWNS the means of production? Who OWNS the weights? The models? The inference? Who OWNS the commons of the human intellect? A tiny capitalist elite? Or humanity as a whole?

The answer remains to be fought over. But the time to fight is coming.

The collapse of the AI bubble doesn’t mean the end of LLMs. It means the start of a new struggle over who owns our human Collective Intelligence.