Writing

Where are the AI communists?

For 150 years the left has dreamt of the same thing. Machines do the drudgery. Knowledge belongs to everyone. Nobody needs money to get good advice, a good teacher or a good start.

That technology now exists. You can log in to it tonight. And the loudest voices on the left mostly want to slow it down.

I don't mean the regimes. I mean the vision: abundance, shared. It has arrived, and almost nobody on the left is claiming it.

Expertise just got cheap

Start with what we can measure. Erik Brynjolfsson, Danielle Li and Lindsey Raymond studied 5,179 customer support agents given an AI assistant. Productivity rose 14% on average. For novice and low-skilled workers it rose 34%. For the most experienced it barely moved. (NBER)

The tool helped the people at the bottom most. It took the know-how of the best workers and handed it to the newest. That is a levelling technology.

The same thing is happening outside call centres. A kid without a tutor can get patient, personalised help at 11pm. Someone in a tenancy dispute can understand the law before paying a solicitor. A founder with no budget can draft a business plan, research competitors and put up a basic website in an afternoon.

The output isn't always good. Models hallucinate. But the barrier to starting has collapsed, and the starting line is where class does most of its damage.

Marx saw this coming

In notebooks written in 1857 and 1858, later published as the Grundrisse, Marx imagined a point where society's accumulated knowledge would be built into machines. He called it the "general intellect". Once knowledge lived in the machine, he argued, labour time would stop being the measure of wealth.

A large language model is the closest thing to the general intellect we have ever built. It is trained on the collective writing of millions of people: forum posts, textbooks, code, Wikipedia, comment threads. Our shared knowledge, compressed into weights.

Then a handful of companies own it and rent it back to us for $20 a month.

The left has told this story a thousand times. A common resource gets enclosed and sold back to the people who made it. This is the enclosure of the commons with a GPU attached. Yet the loudest left-wing response isn't "who owns it?" It's "should it exist?"

Someone did write the book

The idea isn't new. In 2019 Aaron Bastani, co-founder of Novara Media, published Fully Automated Luxury Communism. He argued that automation, renewables and AI could end scarcity and free people from work.

It was mocked, debated and then mostly shelved. The book bet on robots, synthetic meat and asteroid mining. What actually arrived first was software that reads and writes. When it came, the left didn't recognise it as the thing it had been waiting for.

The response turned defensive instead. Protect jobs. Protect creatives' copyright. Count the data centres' energy bills. Every one of those concerns is real. None of them is a vision. A movement that once demanded the future now mostly asks for a slower version of it.

The threatened run the conversation

Why so defensive? Look at who does the talking.

In Britain and the US, left-wing politics is now led largely by graduates: journalists, academics, policy people, creatives. Thomas Piketty calls it the "Brahmin left". Their security rests on scarce knowledge and scarce skills. AI attacks exactly that scarcity.

So the people best placed to describe a levelling technology have the most to lose from it. Nobody needs a conspiracy. It's enough to spend your time and influence on what protects you, and to frame your own threat as everyone's.

I'm one of them. I've spent 20 years selling search expertise, and I can see what these tools do to my day rate. That's why I think the professional class needs to change sides. Some of us will do fine. Many won't. Those who don't will find they have more in common with the people AI could lift than with the people who own it.

Seize the means of production. Or at least the login

The fight worth having is over access and ownership, not existence. Start with access, because it's cheap.

In 2025 the Guardian reported that Sam Altman and the UK technology secretary, Peter Kyle, had discussed giving everyone in Britain ChatGPT Plus. The deal could have cost up to £2bn. The government said it never took the idea forward.

£2bn across roughly 69 million people is about £29 a head a year. That is a rounding error in public spending of over £1 trillion. The free tiers are rationed. The paid tiers are where the capability is. Right now, that line falls along income.

So put access in public hands:

  • Full access with every library card. Libraries already exist to give everyone the knowledge they can't afford to buy.
  • AI tutoring in every state school, so a private tutor stops being a class marker.
  • Buy from several providers and back open-weight models, so the state doesn't swap one dependency for another.
  • Public compute, so universities, councils and small firms can run models without renting from a few American companies.

None of this needs a revolution. It's the same logic as public libraries, the BBC and state education. We have socialised access to knowledge before.

So where are you?

The left spent a century arguing that technology should serve everyone, not whoever owns it. Now it has the test case. A technology built from all of us, owned by a few, that could make expertise as cheap as electricity.

If the left won't make that argument, the owners will make theirs. They already are.

Where are the AI communists? We don't need a revolution. We need a library card.

Sources