{"id":104582,"date":"2025-08-31T07:37:33","date_gmt":"2025-08-31T07:37:33","guid":{"rendered":"https:\/\/www.newsbeep.com\/uk\/104582\/"},"modified":"2025-08-31T07:37:33","modified_gmt":"2025-08-31T07:37:33","slug":"ai-is-like-a-lunch-theres-no-such-thing-as-a-free-one","status":"publish","type":"post","link":"https:\/\/www.newsbeep.com\/uk\/104582\/","title":{"rendered":"AI is like a lunch \u2013 there\u2019s no such thing as a free one"},"content":{"rendered":"

This is a great time to be addicted to AI\ufeff! Everywhere you look, we\u2019re spoiled for choice \u2013 especially of LLMs (large language models) \ufeffsuch as Claude, Gemini, ChatGPT\ufeff. At the last count \ufeff, there were at least 44 LLMs<\/a> publicly available (and doubtless countless others under corporate wraps).<\/p>\n

And most of them appear to be free to use; \ufeffunless you happen to be accessing them via an application programming interface, or API\ufeff (in which case you\u2019ll be paying a fee). Which is odd when you come to think of it, given that LLMs are computationally expensive to run \u2013 and to create. Asking ChatGPT for \u201creliable guides to the history of the\u00a0Crimean \ufeffwar\u201d, for example, requires significantly more computation \u2013 and therefore energy\u00a0\u2013 than putting the same query to Google.<\/p>\n

At the moment, the predominant\ufeff business model used by AI companies is the \u201cfreemium\u201d one: basic use is free, but if you want bells and whistles, you\u2019ll have to fork out. This works for some (including a number of my colleagues, who are paying $200 a month to OpenAI and its ilk), but it\u2019s of no interest at all to the average punter who has always believed that \u2013 as the ancient \ufeffnetizen saying went \u2013 \u201cinformation wants to be free\u201d and, accordingly, is currently enjoying the superabundance of \u201cfree\u201d AI.<\/p>\n

Which raises an interesting question: since there\u2019s no such thing as a free lunch, what\u2019s going on? And where is it heading? As it happens, we know the answer to that last question, because we\u2019ve seen this movie before. Its opening scene was shot in the early days of the \ufeffweb, when people were trying to build businesses on top of the platform created \u2013 and generously donated to the world \u2013 by Tim Berners-Lee (whom God preserve).<\/a><\/p>\n

The more entrepreneurial geeks of the time (the late-1990s onwards) realised that the key to building a successful platform was to \u201c\ufeffget \ufeffbig \ufefffast\u201d, \ufeffensuring that your platform became so large that network effects\ufeff<\/a> would kick in before competitors got their act together. And since the quickest way to get big fast was to provide free services, that\u2019s what Facebook, Google \ufeffet al\ufeff did.<\/p>\n

The current AI arms race between tech giants looks awfully like a death\ufeff march in search of supremacy<\/p>\n

But they also cunningly designed their platforms to make it as difficult as possible for users to leave. With social media, that turned out to be a very powerful disincentive: leavers wouldn\u2019t be able to take their \u201csocial graph\u201d \u2013 their network of friends \u2013 with them. Users feared the resulting\ufeff Fomo, which, coincidentally, is still one of the reasons people find it so hard to come off X\ufeff. Network effects can be tyrannical when push comes to\u00a0shove. Network effects can be tyrannical when push comes to shove online.<\/p>\n

As we swan around in this new, idyllic non-marketplace where Claude, ChatGPT, Gemini, Perplexity, DeepSeek, Llama \ufeffand more flaunt their free credentials and formidable capabilities, two awkward questions come to mind. How long will this phase \u2013 which is clearly unsustainable \u2013 last? And how will the AI industry evolve after the bubble has burst?<\/p>\n

Here\u2019s where the history of the last three decades may again be helpful. The current AI arms race between tech giants looks awfully like a death\ufeff march in search of supremacy. The goal is to get biggest fastest\ufeff and to create \u201cone LLM to rule them all\u201d; a system that does to its competitors \ufeff what Facebook did to MySpace\ufeff\ufeff<\/a> or, in an earlier age, Microsoft did to Netscape<\/a>.<\/p>\n

Whatever happens, though, it looks as \ufeffif LLMs \ufeffmay eventually eclipse not only search engines but social media platforms; and, again, the history of the web can provide some insight into what\u2019s likely to happen when and if that seismic transformation happens.<\/p>\n

In the 1990s, the web was exciting but chaotic until Google arrived and the world changed overnight. But \ufeffthe company lacked a business model until its founders realised that monitoring their users\u2019 clickstreams could provide an information flow that could be fabulously monetised. Thus was born what came to be called \ufeff\u201csurveillance \ufeffcapitalism\u201d \ufeff, which became \ufeff\u2013 and remains\ufeff \u2013 the web\u2019s dominant business model\ufeff.<\/p>\n

One of the great surprises of Google\u2019s monitoring was how much clickstreams can tell you about people\ufeff. But we know already that people\u2019s interactions with LLMs can be much more revealing<\/a>, more intimate and surprising than their clickstreams. Just after I wrote that sentence, a New York Times article \ufeffpopped into my inbox. The title: \ufeff\u201cWhat my daughter told ChatGPT before she took her life.\u201d\ufeff\ufeff<\/p>\n

So if AI corporations start surveilling their users\u2019 interactions with LLMs \u2013 as they eventually will, just as social media companies did \u2013 they will have inadvertently discovered a new, and even more pernicious, business model. Let\u2019s call it \u201c\ufeffthoughtstream \ufeffcapitalism\u201d.<\/p>\n

Photograph by Karjean Levine\/Getty<\/p>\n","protected":false},"excerpt":{"rendered":"This is a great time to be addicted to AI\ufeff! Everywhere you look, we\u2019re spoiled for choice \u2013…\n","protected":false},"author":2,"featured_media":104583,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[20],"tags":[554,733,4308,86,56,54,55],"class_list":["post-104582","post","type-post","status-publish","format-standard","has-post-thumbnail","category-artificial-intelligence","tag-ai","tag-artificial-intelligence","tag-artificialintelligence","tag-technology","tag-uk","tag-united-kingdom","tag-unitedkingdom"],"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/posts\/104582","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/comments?post=104582"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/posts\/104582\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/media\/104583"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/media?parent=104582"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/categories?post=104582"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/tags?post=104582"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}