What if AI's biggest innovation is not intelligence at all?
What if its real achievement is finding a way to take other people's work, package it, and charge them to access it?
In this video, I argue that the economic model behind AI deserves much closer scrutiny than it currently receives.
AI systems are trained on enormous quantities of human-created content: books, articles, websites, blogs, videos, images, music and much more. Without that content, AI would have nothing to learn from and nothing useful to say.
Yet the creators whose work makes these systems possible are rarely paid. In many cases, they are not even asked for permission. Their knowledge, creativity and labour become the raw material for products that are then sold back to the public.
That raises fundamental questions about copyright, ownership, economic justice and the concentration of wealth.
Drawing on ideas from political economy, I argue that what many AI companies are doing is best understood as rent extraction. They are creating systems that allow them to charge for access to information that they did not originally create. The result could be a new form of digital landlordism, in which a handful of technology companies gain the power to charge rent for access to knowledge itself.
I also explain why this issue matters to me personally. Over more than twenty years, I have written tens of thousands of blog posts, produced more than a thousand videos and written books whose contents now appear to have been absorbed into AI systems. Like countless other creators, I have never been paid for that contribution, even though it helps make these systems valuable.
The debate does not stop with copyright. It extends to democracy, regulation, taxation and the future distribution of economic power. Who should benefit from AI? The creators whose work made it possible? The public? Or the shareholders of a small number of giant technology companies?
I also examine the wider political implications of AI, including the very different perspectives offered by Tony Blair and Pope Leo XIV. One sees AI as a central pillar of our economic future. The other warns of the dangers that arise when powerful technologies are allowed to operate without sufficient democratic accountability.
This is not simply a debate about technology. It is a debate about economics, democracy, ownership and power. It is about who gains, who loses, and who will control one of the most important technologies of our age.
If AI is going to shape our future, then we need to decide whose future it will be. That is the question at the heart of this video.
This is the audio version:
The Debate Ammunition for this video is available here.
This is the transcript:
AI is an economic con-trick, and it's time to say so, in my opinion. Large language models do one thing, and they do it quite well. They take vast amounts of other people's data without paying for it, and then they process that data and sell access to it. That's what the AI business model is.
The programming makes all that output seem reasonable and well-constructed, although whether it's right is another matter because we all know that garbage in, garbage out still applies to AI content, and AI admits it and charges us nonetheless for the content that it supplies, but this does not avoid the fact that there is a con-trick going on, which is that AI is using other people's data, and it isn't paying for it. That's my concern.
But saying that, I won't pretend that AI has no value; it does do some things really genuinely well. It can assemble and organise information at extraordinary speed, faster than anything before it.
And like search engines, which came before it, it gives us access to information in new ways, and that is useful, I'm not going to pretend otherwise.
But it goes further than search. It narrates, organises and draws apparent conclusions, and that is a real capability, and I will not deny it, but the capability does not justify the AI business model. That is my argument.
And I can speak personally here. I am one of the victims of this process. Over 20 years, I have written more than 25,000 blog posts and have made more than 1,000 videos. The transcripts of those videos are freely available on the web, and I've also written books and all of them, I suspect, are scraped by AI, and I know that because I can get AI to write a summary of them. It does, therefore, know what my books are about.
Has any AI company ever compensated me for that raw material? No, is the plain, straightforward answer.
Have they tried to? No.
Am I expecting them to? No. And let's be clear, despite that, they now wish to charge me to access that information of mine, which they have now put onto the web.
Is this a breach of my copyright? I think it probably is. I publish under a Creative Commons licence that says my material is free for anyone to use, but not for commercial purposes. Then I insist that people must come to me and ask for permission and we might agree a fee, but these companies have used my work with the clear intention of making a profit. I've not given them permission to do that. These materials are not licensed for use in that way. In that case, I could argue that my copyright and my licence conditions have been breached by these companies.
Legal cases are, I know, already being brought against AI companies for this reason by a whole range of authors. I have been invited to join some of those cases. I haven't so far, but I'm thinking about it, and I do support those cases. The principle at stake here matters enormously.
And what AI companies are doing has a precise name in political economy. Let's be clear about what it is. It is rent extraction.
A rent in political economic terms is an access fee charged for an asset by someone who claims the right to own it. Nobody asked these companies to scrape the web. They did so voluntarily, and having assembled databases from other people's work, they now charge subscriptions to access them. That subscription charge is a rent, nothing more and nothing less.
We are paying to access what AI companies have assembled, often it seems without permission, because they haven't asked for mine. And there has always been a fundamental problem with rents, and that is who actually gets paid when a rent is settled.
When you pay rent on land, for example, when you rent a property, the land gets nothing. Let's be clear about that. You can't pay a plot of land for the right to reside on it. That's not possible in this world, but the landlord does get your payment. The same logic applies precisely to AI.
The people who created the content that populates the databases that AI created get nothing from their rent. They stand in the same position as the land itself in the rent of property. The asset creator receives no reward at all. Those who ring-fenced and enclosed the property collect the entire return.
'Twas always thus, as students of history well know. This is the principle of enclosures, which changed, for example, the history of the UK in the 17th and 18th centuries, and AI tech companies are, in this sense, landlords. That is precisely and literally what they are. They make returns that they have not truly earned in full.
I'm not disputing that they make part of the property that they sell. That I can't dispute, and I'm not saying they shouldn't be paid for it. But I am saying that the underlying data they sell is not apparently theirs to charge for, but they can because the law permits them to do so. And law, as ever, is written by the rich for the benefit of the rich.
That was how so much of the common land of England was enclosed at the time I've just mentioned, the 18th century. That is exactly what is happening with AI regulation right now as well.
AI creates the same economic distortions that land ownership has created throughout history. Landlordism has always concentrated power in the hands of the few, and so will AI.
We can see this playing out in live political debates right now.
Tony Blair has written an essay in the course of the last week, proposing AI as our economic future, and we're making a video about that. If it's out by the time this one is, we'll put a link in down below. If not, watch out for it. Pope Leo XIV has issued an encyclical raising profound concerns about AI, and I have made a video on that one. There is a link down below, and what Blair wants and what the Pope resists represent fundamental conflicts about who benefits from AI.
Blair's vision means a massive redistribution of wealth away from ordinary people towards the owners of AI. And as a consequence, he wants to benefit a very limited number of people in the world who will hold the rights to impose these rents, and we will have to pay them.
The distortion in economic well-being that this could produce may be without precedent in economic history. A tiny number of AI tech companies will hold the right to charge rents on access to knowledge. Everyone, individuals, businesses, public services, governments, will have to pay those rents.
I've long debated the inequality arising from tax injustice, from financialisation, from wage suppression, and from the destruction of the social safety net. But the inequality that unchecked AI rent extraction could produce may dwarf all of that. This is the oldest story in economic history, now playing out at digital speed and global scale. And that is why the Pope and I are worried about it.
As the Pope says, we need to decide who truly governs AI. It must be made democratically accountable and not just to shareholders. It must pay for the property rights of others it has used without permission, and its content must be regulated so it cannot abuse, mislead, or cause harm.
This is not a technical question. It is a profoundly political one, and it is urgent. The window to act is not indefinitely open. AI charges must be regulated. Excess profit extraction from a captive public cannot be permitted. AI profits must be taxed appropriately and fairly, and the taxation must apply where the impact of AI arises, not where its file service happens to be located, which may well turn out to be in tax havens. That's been the pattern to date. The same tax justice principles that apply to every other form of extraction should apply here. Nothing about AI exempts it from paying its fair share of tax.
We have to recognise that what is going on is, in most cases, not a process of genuine information generation or innovation. There's a bit of that, but most of the time, what is going on is a process of information extraction from others. That is the thing that we need to deal with.
The politics of care is focused upon the interests of people, including creators, and a politics of care focused upon the interests of people would necessarily demand that we take action to manage AI, to ensure that its capacity is used for the benefit of human well-being.
We all have the agency to act on this issue. I can act, you can act, we can all demand that the government act. You have the ability, as I do, to demand action on this, and we must use that right now because if we don't, we're all going to pay a price in the future. That is something I would regret.
That's my opinion. What's yours? You have a right to disagree, of course. As the AI companies do. They can come on here and tell me why I'm wrong. But when doing so, if you'd like to fill in the poll below, that would be good. Let us have your comment. Like this video if you do. Please do share it, and if you're so inclined and want to buy us a coffee so that we might continue to make videos like this, we'd be really grateful.
Poll
Loading ...
Thanks for reading this post.
You can share this post on social media of your choice by clicking these icons:
There are links to this blog's glossary in the above post that explain technical terms used in it. Follow them for more explanations.
You can subscribe to this blog's daily email here.
And if you would like to support this blog you can, here:
Buy me a coffee!
[…] AI Is an Economic Con-Trick. The video relating to this content is here. […]
Let’s be clear we are talking about one form of AI here – LLM’s. I’d never pay for them, for my purposes I just use Google, that gives a good overview. If I want a deeper dive I wouldn’t use an LLM, I’d go straight to the source. Example: I sent a link of your “Richard murphy on green new deal” to one of my WhatsApp groups. Someone asked me about rent extraction. I said for an overview use Google or an LLM for a bit more but if you are really serious go read Mazucatto. AI will bring more casual readers in. The serious, the geeks like me will always pay for detailed content.
Good point, Kevan.
I wonder, though, how sustainable is this approach for a future where the key holders to all available knowledge might be the owners of AI technology.
My family sometimes tell me I have an evil imagination…
If I had the resources, and wanted to poison the AI food chain, so it filled up with unreliable garbage, I would become a publisher of maliciously edited garbage, ensure it was well optimised for scraping (what they used to call SEO) and then employ a bot army to interrogate my created (garbage) content as paying customers of my chosen opponent’s AI product.
But that would be very evil. It might even be possible. Do not do it.
But in a world with some of very evil very powerful, even state level actors and including some very wealthy digital techbros with skin in the game) it might even happen. “AI Wars”???
The “publishing garbage” bit is already well under way, consult your favourite MSM website for suitable material.
Maybe my family are right, I do have an evil imagination. I do not condone this activity, although judging by the amount of “AI slop” around, maybe some evil person is already doing it?
People have started “data poisoning” to fight back against this. It’s definitely justified IMHO. It’s well covered on YT so I won’t waste your time explaining.
I have seen it
There ha been a recent very notable case in medicine – an entirely made up disease deliberately placed to be propogated – and it was
The concept of data poisoning is new to me, and I feel somewhat encouraged by it. This article gives a good overview, including touching on the ethical and legal implications.
https://theconversation.com/in-the-face-of-rampant-ai-is-data-poisoning-a-new-form-of-civil-disobedience-280146
Excellent summary, thanks for the link
All fair points. However. There is a weakness within A.I. stemming from its use of other peoples work/documents. Readers are invited to ask one of the platforms this question: “Give me a summary of European Electricity Market function and the underpinning rationale”. You will get the usual contradictory nonsesnse. Or “What do you think of European Eelectricity markets and their functioning”. Trash in Trash out. I spent some time training an A.I. (yes – I pay for it) on this subject. What it produces now is high-speed analyses/demolition of the self-serving garbage the EU produces in this area (it was sad as I pointed out the internal contradictions in what it produced – leading to “reverse ferrets” – of course “it” did not produce those contraditions – they were inherent in the docs it accessed – but its logic was able to “see” this – once it was pointed out).
It also saves me time reading overly-long work (e.g. the Spanish Transmission operator (TSO) produced a 40 page report on the 2025 Spanish outage. I read it & arrived at conclusions. ENTSO-E – the TSO Euro, their report was 400 pages. I point blank refuse to read a report on the same subject that could be covered in 40 pages . I passed it all to the A.I. – read them both, identify differences, summarise. The summary converged with my own. A.I. = tool. saves my time & keeps my blood pressure low = avoids me having to read self-serving trash. A.I. governance? Most definitely not with the tech-bros (robber barons of yore).
We make similar use of AI
Thats broadly my conclusion, as someone tracking AI over many years and trying to shape party policy. Good for summarising lengthy documents on topics you understand and hence can check. With tightly focused questions – thank you Richard. Preferably targeting material that would be public domain.
I have seen at least one company which had a clear statement that they would not use creative material belonging to others but Id not rely on all companies behaving that way in practice. The biggest concern has to be that so much of AI is in the hands of people who are demonstrably deeply unethical. And Im seeing corporate and financial people rubbing their hands with glee at the prospect of the people they can get rid off, leaving the problem to the state that they despise so much.
Thanks
Agree.
As said theft is theft. Bring these robber barons to justice.
I am not saying it is theft
I am saying it is wrong
As Kevan says, this is about one form of AI – large language models – which are the flavour of the month currently. They are transformers using artificial neural networks (which are nothing new – a colleague was using simple versions of them for scientific purposes when I was doing postgraduate research in the early 1990) – layers of nodes connected by links with different weights – that take in a heap of data, convert that data into tokens connected by a contextual model – the nodes and the weights between the nodes – and then respond to a prompt by providing a sequence of tokens based on that statistical model that connects the previous token to the one that most likely comes next. It is remarkably simple and remarkably powerful. The original LLMs were just that – a model created by linguists to parametrise and analyse language, to break it down into syntactic and semantic building blocks. They were never intended to show “intelligence”. And really they are not intelligent. They have no understanding of how the world is or should be, or what is right or wrong. They are just rather neat and somewhat useful statistical models.
But there are many other forms of AI for different applications, for data analysis or search, or for manufacturing. LMMs are popular so they are being pressed into many uses where they are plainly unsuitable.
Your thesis is entirely plausible. AI is the latest iteration of the lack of double entry book keeping in capitalism that you identified some time ago.
The American film maker Michael Moore says in his film ‘Capitalism: A Love Story’ ‘Capitalism is system of giving and taking……..mostly taking’.
And your post underlies something I keep try to ram home – inequality is a factor of production from monopolism – the desire – Pleonexia – to dominate, to have/covet what others have. Neo-liberalism justifies and celebrates the worst in human beings, our most anti-social traits and turns them into virtues.
AI cloaks itself in our desire for agency, but really is an agency for the market and not us. And we have to remember that that is all it does really – monitor and learn from what we buy or look for . And how rational are human beings when they are buying stuff – even railway books as well (sorry)!! So it is all very well saying AI learns from what we do, but what and who is AI looking at? Who provides its eyes?
The actual reward sold using AI seems to be more leisure time for society but really is just an exercise in reducing labour costs and delivering more profit for organisations that fund what is a political cartel (STP) leaving the state with jobless people. This cartel insists on people seeking work, does not invest in jobs, does not seek to tax capital on its spillover effects, offers less social protection every year it is in office and has no idea how this leisure time will be funded except through unsustainable personal debt/credit. ‘Fucked up’ is not even an adequate description of this situation.
These LLM AIs are based upon the Silicon Valley behaviour of “Move Fast And Break Things”.
In my opinion this beaviour is the absolute core of the problem, needs to be stopped and then replaced by TIARA.
Governments could do this easily.
They’re the ones in charge, it’s not the LLM AI companies.
E.g. the US-Government could plain simply say:
“Well, we don’t like Grok / OpenAI / Antrophic / … anymore so it’ll be shut down today”
But therefore they would need to read your article about fear and accept uncertainty.
Their Fear Of Missing Out is driving them at the moment which gives these companies the upper hand.
Thanks Richard , excellent and very timely.
The AI behemoths have so deep pockets that they can influence decision making, the moment any country attempts to reign them in. I would assume they have to face the full wrath of US govt (they are the cheerleaders of this AI boom, yes – China catching up).
Do you think EU / UK can regulate ?
They could
That is not the question
Will they is
No is the answer
Richard you may want to look at Press Gazette and search for court battles re copyright with AI companies.
We journalists are trying to manage a system that is taking our content without paying, and as the big social media companies are simultaneously switching off our links to readers, the news industry (and most of us are honest regional journalists scrutinising and upholding the foundations if democracy) is in existential danger.
It really is serious. Who will investigate and report on our country, it’s politicians, crooks, justice systems. Services, councils etc if we are destroyed?
An excellent piece of work Richard.
In college ..many years ago ..studying BSc in Multimedia and Information studies… we were told that when producing an essay we HAD to give references .there was a specific format that we had to adhere to..and in some cases ask permission to use the information provided by another. It’s only fair ..using someone else’s work that they had spent time researching and producing. If you use another persons work ..it is plagiarism…no ifs or buts..you are stealing another person’s work. And who knows how you might twist and change it slightly for your own nefarious purposes…things have changed it seems….
As Yanis Varoufakis has suggested, we are now in an age of Technofeudalism.
I can recommend Emily Bender and Alex Hanna’s book “The AI Con”
https://www.hive.co.uk/Product/Emily-M-Bender/The-AI-Con–How-to-Fight-Big-Techs-Hype-and-Create-the-Future-We-Want/30554961
They make a clear case for the wholly immoral, if not also illegal, actions of the AI industry.