The commons that AI steals from: treat it as a public good?
by Mariana Mazzucato
AI Should Help Fund Creative Labor
Do click through the link if you are concerned, as I am, at the way LLMs have created hugely capitalized enterprises on the nonconsensual use of the words and images of millions of creators. Decrying the fait accompli doesn't seem to me to be enough to accomplish mitigation. Mazzacutto is, in my opinion, the most interesting economist in the public sphere. Although it seems unlikely in the current political atmosphere, treating the aggregate online commons as a public good is an interesting proposal
Behind every AI-generated response lurks a vast, invisible workforce – writers, singers, journalists, poets, coders, illustrators, photographers, and filmmakers – whose creations have been used without permission or compensation. These creators have never met, let alone billed, the Silicon Valley titans profiting from their labor.
...
Rather than patching up a market ill-suited to managing public goods, governments should actively nurture the cultural commons by steering innovation toward public purposes. Just as we pool taxes to fund streetlights, law enforcement, and basic research, the production of creative content in the era of generative AI should be publicly supported, and its outputs kept in the public domain. In short, the state must be entrepreneurial. This idea is not new. The BBC license fee, France’s National Center for Cinema, and even US states’ film-production tax credits have long supported major global hits, from the BBC documentary Blue Planet II – which was reportedly watched by 80 million people in China and temporarily slowed the country’s internet – to the ABC/BBC cartoon Bluey, which became the most-streamed program in the United States in 2024. Above all, the public model generates immense value. It provides creators with stable funding, fosters innovation aimed at citizens rather than advertisers, and enables artistic risk-taking and experimentation. It also helps preserve shared cultural heritage, in turn enriching education, strengthening social bonds, and galvanizing democratic debate in ways that market-driven models rarely do. Of course, the case for such an approach extends far beyond broadcasting and cinema and applies to all forms of art, media, and creative expression. Because generative AI models are trained on human-created content, the value of art to society takes on a new dimension. By increasing the volume and diversity of creative output, these technologies amplify the reach and impact of human creativity. One could argue that by repurposing creative works, AI has expanded the art multiplier: each dollar spent on the arts now yields its usual social return, as well as additional value derived from its incorporation into AI systems.