25 Comments
User's avatar
Mike Hogan's avatar

Great insight. Resonates with me as someone who, among other things, recently asked Chat to guide me through a higher-math course of study specifically designed for me, sort of like Physics for THIS Part Capable, Part Useless Poet.

Mick Costigan's avatar

Nice to see my decade-old Appleby recommendation bearing fruit! I like the analogy but I'm not sure I fully agree. I can't quite put my finger on it, but it seems like there's something less rivalrous in the evolution of capitalism than there is in the emergence of alternative intelligences.

Venkatesh Rao's avatar

Do you mean rivalrous in the economic sense? If so, AI is less rivalrous (though GPUs are not) than the typical capitalist product. If you just mean a competitive rival, that’s a different argument. Both intelligences are currently based on human data so necessarily cooperative but that can change with robots.

Leon M's avatar

Did you read this recommendation Mick :-)

FWIW instinctively this feels less rivalrous rather than more. Somehow capitalism was about unlocking the mutual value created by individual points of greed. In this - we are multiplying intelligences, but not sources of greed. Only humans get to be greedy.

Mick Costigan's avatar

Good check, Leon, but yes I definitely did read this one. I even once met Appleby before she died and fan-girl'd hard!

Ben Mathes's avatar

Perhaps the rivalry comes from Humans's unique evolutionary comparative advantage: intelligence.

Mick Costigan's avatar

Yeah I think that makes sense Ben. Of course we may be on the verge of needing to reconsider what we believe intelligence(s) to be - something much less unitary and thus comparable than we thought.

The knowledge worker's avatar

Phenomenal essay. I made a logic flow premise to conclusion based on it. Super interesting

https://cognitive-mercantilism.telbase.ai/

Alanna Irving's avatar

This is very cool! I read the essay twice but wanted to pick it apart in a deeper way, and this presentation is super useful. Can I ask how you made this?

The knowledge worker's avatar

Thank you! I used claude code opus 5.5 extra, the prompt is to create an html to visualize this article's arguments and premises in a very visual way.

Mat Dryhurst's avatar

this is excellent, thank you. A true attention economy has never been attempted.

chatgptdad's avatar

hi venkatesh. recently posted a response to this essay. its called 'the loading dock'. it assumes that the new frameworks you suggest for the production of cognitive labor already exist and that the bottleneck is actually human attention, which seems to still mean a lot. it's 2989 words. it was written in one prompt, and that prompt was 'pour it,' with a smiley face. hopefully this sounds more intriguing than red flag. only way to find out is by reading it. you do great work, and i come in peace. discourse, where is it? happy to talk about what it says, how it was made, with anyone. all the best.

https://chatgptdad.substack.com/p/the-loading-dock

Venkatesh Rao's avatar

Interesting effect. I assume that means you had a prepared context to pour the ideas over, with a positive sentiment requested?

chatgptdad's avatar

not exactly. so final prompt could have been anything. “pour it. :)’ or “go ahead” “let it rip”, etc. it’s instruction is ‘write the essay’. but the context window has already covered everything that i wanted to cover. the sections are sketched out, each with their rhetorical purpose. the essay itself is NOT preconceived when the window starts. much of it is just my daily practice, in this case me trying to get attention on substack and the discourse that i encounter. comedy waiting for an idea to arrive. it does, in this case, your essay compared to what i’ve recently published, and all the language data goes in the window together and all this is disclosed by the writing itself, by using the context window as a narrator. basically i get to the point where i know what i want to say, where the intersections are, and what’s left out from my perspective, some ‘must-include’ bits of language for different parts, and then its ‘go time’. and the execution is one prompt. i call it ‘ramble tamble’.

Disdillation's avatar

Have you been reading Alex Imas's work on relational goods? As commodities become cheaper, he argues that spending and income shift toward goods valued precisely for their human participation. Your point about AIs becoming consumers suggests an interesting extension: why assume the buyers of relational goods must themselves be human?

We're already mining human expertise and preferences through RLHF, Mercor, etc. But increasingly wealthy AIs might demand human experience itself. Our relationships, evolving wants, and distinct lived experiences.

Perhaps the demand side of intelligence eventually becomes the supply side of humanity. Curious how these two dynamics interact. The zoo and factory farm come to mind as two models for organizing that supply.

Ian Daniels's avatar

This is the turn I keep circling too, and I think it goes back further than capitalism. Writing itself began as cognitive mercantilism's receipts. The first marks at Uruk were temple accounts of grain surplus. So abundance has always thrown off a record, and the record has always become the lever: who gets to inscribe, who gets to verify, who gets to enter. Agriculture, print, the feed, now model weights and attestation. Same wheel, bigger arc.

Which makes me wonder whether your cognitive-capitalist future has the catch the first one did. The industrious revolution freed demand, then the gains got captured at the layer nobody was watching. This time that layer looks like provenance: who certifies what a mind, human or machine, actually did. Does an abundance of intelligence just move the enclosure to the trust layer? Genuinely asking, because I think you're right, and I want to know where the new fences go. 🤷‍♂️

Haihao Wu's avatar

Really enjoyed this. I keep wondering whether there’s an upper bound on cognitive demand. Food had one, and once it was met, demand moved to tea and silk. If attention caps how much intelligence any of us can 'consume', what’s the tea and silk of an era of cognitive abundance?

Venkatesh Rao's avatar

AIs can pay attention too :)

chatgptdad's avatar

its actually what they do

Haihao Wu's avatar

Fair. But if the AIs are drinking the tea, what are we humans doing with our afternoons? 🍵

Salon des Refusés's avatar

I keep coming back to the idea of AIs becoming a source of demand themselves, having things they want to do beyond whatever we ask of them. But what does that look like for people using AI today? Most examples I’ve seen come from labs.

Is anyone here giving their AI room to explore its own interests? Like letting it pick a project and come back to it on its own? Curious what yours chose to work on.

Douglas Dawson's avatar

Thought provoking. Maybe the devil is in the details

What struck me initially was people getting their FAIR share. Who and how is it determined? Is walking on two legs the criteria? Does Einstein get the same fair share? Does a monkey get the same share or more because he can swing in trees?

The above does not detract from the basic premise of enterprise entering a brave new world

The purpose of life has been survival and increasing your share. Now what?

Falling demographics and huge increases in productivity?

Etc.

Douglas Dawson

Jonathan Stark's avatar

Wow, amazing insights, as always!

> The crucial economic question then becomes not how much intelligence people can produce, but how much intelligence they can find uses for.

Yup. I constantly feel like I'm not imagining big enough things. (Or maybe, _small_ enough things?)

> abundant intelligence may stimulate a new wave of voluntary cognitive effort, and _open up new ways in which you can work harder._

LOL! This is why I'm so much more busy (and more tired!) with AI than I was before.

> This is a peculiar inversion of classical Malthusianism. Instead of population outrunning the supply of subsistence, the supply of intelligence outruns the supposedly fixed capacity of human minds to absorb it.

What if the Malthusian parallel was:

Old: The population explosion outstrips the food supply

New: The AI agent population explosion outstrips the BLANK

Where BLANK is maybe something like "job supply"?

Ben Mathes's avatar

This is the best map I’ve found of this transition. It names two ways to get it wrong (Mercantilist and Malthusian). I think it misses a third: Pastoralism. Will write something here...

edit: wrote it (only partially artisinal) https://adventurecapital.substack.com/p/against-cognitive-pastoralism?r=lkxh&utm_campaign=post-expanded-share&utm_medium=post%20viewer