I keep thinking about AI in a slightly unfashionable way.<\/p>\n\n\n\n
That one of its economic functions is to let capital access skilled intelligence more cheaply, without returning quite as much of the value created to the person who possesses<\/em> the intelligence. Which means, awkwardly, wealth gets better access to intelligence while the people providing it get less access to the wealth it creates.\u00a0<\/p>\n\n\n\n I genuinely don\u2019t seek out these connections. I\u2019ll be casually reading Sally Rooney before bed. Geranium mist in the air. Peppermint tea within arm\u2019s reach. Then my brain says, \u201cOh, you thought we were done for the day?\u201d<\/p>\n\n\n\n This is what I notice. UK companies moving from defined benefit pensions to defined contribution schemes, aka contingent liabilities becoming predictable expenses. Offshoring. Gig workers. Shared service centres. Leasing instead of owning infrastructure. Customer self-service.<\/p>\n\n\n\n Consider ordinary customer experiences: checkouts. Returns. Online check-ins. Banking. Booking. Things that once upon a time required somebody\u2019s paid time now quietly requires ours. But like, unpaid.<\/p>\n\n\n\n All different mechanisms, but the same direction of travel.<\/p>\n\n\n\n There\u2019s always been an incentive for businesses to reduce overheads, make costs more predictable and pay less for what they need. I don\u2019t think there\u2019s a particular villain story here. Mostly, it\u2019s incentives doing what incentives do.<\/p>\n\n\n\n Intelligence has historically been stubbornly attached to a person. You needed the lawyer. The analyst. The programmer. The designer. The person who knew things, noticed things, could reason through things.<\/p>\n\n\n\n And because the intelligence came with a human attached, some of the wealth created by that ability flowed back to the human. <\/p>\n\n\n\n AI changes that bargain. Well… not completely. And maybe not even as dramatically as certain advocates imagine. But enough to make the economics interesting.<\/p>\n\n\n\n If capital can access a meaningful slice of that intelligence without needing the human attached to it, it retains more of the value created. That, to me, feels like one of the more interesting economic propositions hiding underneath AI. \u00a0<\/p>\n\n\n\n I don\u2019t disbelieve<\/em> the human-flourishing proposition. New tech does indeed create wonderful things. But I get sceptical when that\u2019s 90% of the PR narrative. I react the same way when I see a shop selling a \u20ac5 can of Coke. I\u2019m not buying it.<\/p>\n\n\n\n YouTube and Instagram feel like a useful precedent. They created genuine new livelihoods and outsized returns for some people. But they also taught millions of people to spend unpaid hours adapting themselves to an algorithm. Both things are true.<\/p>\n\n\n\n Platforms follow the incentives. And so do the creators… but the creators have nervous systems.<\/p>\n\n\n\n Most people aren\u2019t spending their evenings thinking about inference economics and model architecture.<\/p>\n\n\n\n And, tbh, why would they?<\/p>\n\n\n\n They\u2019re probably using GPT to talk through a breakup, rewrite an awkward email or ask why their boiler is making a weird noise.<\/p>\n\n\n\n Meanwhile, an extraordinary amount of money has been bet on AI becoming economically significant. That money eventually needs a payoff. <\/p>\n\n\n\n So I sometimes wonder how much of the headlines, announcements and predictions reflect widespread public fascination. And how much is simply what it looks like when a very large amount of wealth needs something to become<\/em> important.<\/p>\n\n\n\n OpenAI and Anthropic have distribution, sure. But so do the companies already<\/em> sitting inside our digital lives and corporate infrastructure. <\/p>\n\n\n\n Google. Salesforce. Microsoft. Amazon. On a B2B level, they\u2019re already embedded. And they<\/em> can build AI too.<\/p>\n\n\n\n Meanwhile, cheaper Chinese models keep pushing down the price of inference. So how comfortable a place is there, ultimately, for a standalone model company like OpenAI or Anthropic?<\/p>\n\n\n\n Model scarcity reduces as intelligence gets cheaper. So they have to become useful enough, sticky enough, and embedded deeply enough to survive if intelligence becomes too <\/em>much of a commodity.<\/p>\n\n\n\n Maybe they manage it. Or maybe, over time, the distinction between model company and hyperscaler becomes less interesting. For example, if it just becomes another quiet layer of infrastructure.<\/p>\n\n\n\n Which would be such a comically mundane ending to all of this hysteria.<\/p>\n\n\n\n Billions invested. Endless headlines. Predictions about the end of work, the end of creativity, possibly the end of humanity.<\/p>\n\n\n\n And years from now, the \u201crevolution\u201d is three lines buried in a corporate 10-K.<\/p>\n\n\n\n AI becomes part of the hyperscaler furniture.<\/p>\n","protected":false},"excerpt":{"rendered":" I keep thinking about AI in a slightly unfashionable way. That one of its economic functions is to let capital access skilled intelligence more cheaply, without returning quite as much of the value created to the person who possesses the intelligence. Which means, awkwardly, wealth gets better access to intelligence while the people providing it […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[5],"tags":[],"class_list":["post-951","post","type-post","status-publish","format-standard","hentry","category-thoughts"],"blocksy_meta":{"styles_descriptor":{"styles":{"desktop":"","tablet":"","mobile":""},"google_fonts":[],"version":7}},"_links":{"self":[{"href":"https:\/\/imruls.blog\/index.php?rest_route=\/wp\/v2\/posts\/951","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/imruls.blog\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/imruls.blog\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/imruls.blog\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/imruls.blog\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=951"}],"version-history":[{"count":5,"href":"https:\/\/imruls.blog\/index.php?rest_route=\/wp\/v2\/posts\/951\/revisions"}],"predecessor-version":[{"id":958,"href":"https:\/\/imruls.blog\/index.php?rest_route=\/wp\/v2\/posts\/951\/revisions\/958"}],"wp:attachment":[{"href":"https:\/\/imruls.blog\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=951"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/imruls.blog\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=951"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/imruls.blog\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=951"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}The pattern<\/strong><\/h5>\n\n\n\n
The human attached<\/strong><\/h5>\n\n\n\n
The money behind the noise<\/strong><\/h5>\n\n\n\n
The end state<\/strong><\/h5>\n\n\n\n