The Machine and the Commons
The Machine and the Commons
The Machine and the Commons
The argument against artificial intelligence is strongest when it begins with ownership. These systems did not become powerful by magic. They were trained on writing, images, code, speech, behaviour, research, art, performance, and the ordinary traces people leave behind as they move through the digital world. The machine learned from a common store of human activity, but the infrastructure around it was built as private property.
That is why the language of theft has force. AI does not only automate work. It reorganises the ownership of knowledge. It takes material produced socially, processes it through closed systems, and sells access back to the public. The contribution is broad, but the revenue is narrow. Those with models, chips, cloud contracts, data centres, capital, legal teams, and distribution power sit closest to the gain. Many of the people whose work helped make the systems useful receive little more than the promise that innovation will eventually benefit everyone.
But that is not the whole story.
AI also lowers the cost of explanation. It can help a student understand a court ruling, a worker read a contract, a small business owner draft a letter, a reader enter a difficult book, or an independent writer compare several documents without needing an institution behind them. It can translate technical or bureaucratic language into something usable. It can open the gates around knowledge that were once protected by credentials, money, time, and proximity to elite institutions.
This is the contradiction. AI can democratise knowledge at the point of use while enclosing knowledge at the point of ownership. The same tool that helps more people understand the world can also concentrate the infrastructure through which understanding is produced.
That means the political question is not whether AI is simply theft or liberation. It is both possibility and enclosure, depending on who governs it. Who owns the models? Who controls the data? Who is compensated? Who can inspect the systems? Who has access? Who carries the costs of energy, surveillance, labour displacement, and dependence on a few dominant firms?
A more equal world would need wider access to knowledge. But access without power can become a new form of dependency. The task is not to reject the machine as such. It is to decide whether the knowledge commons will be enclosed by capital, or organised in ways that return some of its power to the people who made it possible.