Commons over, baby

Is AI destroying our digital commons? Can we stop it? Should we?

I wrote recently about the technologisation of culture, and how we needed to resist – as the Luddites did – the introduction of machinery “hurtful to commonality”. Instead we should prioritise the health of human culture and of our collective digital “commons”. We should not allow technology to destroy either the creative industries or the incredible open-access wealth of knowledge that is the web.

Clearly the gravest technological threat to that common culture at the moment is AI.

What are the practical ways of ensuring that the commons is protected in the face of growing AI capability, though? Is there anything we can do? How do we need to think about this stuff? What moral and practical frameworks are there?


Robin Sloan wrote a piece back in February about his views on AI ethics, and what would need to happen for him to conclude that AI was either ethical or unethical.

He starts with the undeniable observation that AIs have consumed lots of text and images from the internet; they’ve absorbed our digital commons, with no regard for whether it was rights-protected or public access, duplicating it all into their training data. Only by consuming this quantity and range of information has it been possible to build LLMs that are potentially useful; training them on public domain works doesn’t seem to work very well. As Sloan says:

“In fact, the only trove known to produce noteworthy capabilities is: the entire internet, or close enough. The whole browsable commons of human writing. From here on out, we’ll call it Everything, which is short for Everything That Can Be Accessed Online, Plus Some Extra Pirated Stuff, Probably.”

He asks the question: is that okay?

People’s instinctive responses to that question are often fairly tribal, and based on their relationship to that commons. The writers, artists, bloggers and coders whose work has been consumed tend to view the act as a hostile and transgressive one, an act of theft and a violation both of existing intellectual property laws and of the more informal morality that has grown up around them. The tech folk who are building LLMs and applications on top of them tend to view it as a perfectly reasonable process of learning, no different to Google building its search index or a human being watching a film or reading a book.

To Robin this framing isn’t the correct one, and the question isn’t particularly a practical one anyway. AI companies are going to keep slurping up content whatever we do, we don’t have a means to stop them. The horse has bolted right over the horizon already; conversations about stable door maintenance shouldn’t be our focus. Instead, Robin concludes, the question is a moral one, and his conclusion is fundamentally consequentialist. What will the result of these models’ existence be, and do we like the sound of that? What will their impact be upon the commons? Can there ever be an excuse for such wholesale consumption of our shared cultural artefacts?

In Robin’s view, if AIs are used to advance science, to develop new technologies, to push forward the cause of human progress at a faster clip than human ingenuity alone could achieve, then they are obviously moral, even if they end up damaging our current cultural fabric. As Sloan says, “if these machines churn through all media, and then, in their deployment, discover several superconductors and cure all cancers, I’d say, okay… we’re good.”

If on the other hand they’re used to produce creative works that crowd out and supersede those created by humans, without achieving anything else substantial, then to Robin they are obviously immoral.

This strikes me as a reasonable way of thinking. We must protect the commons; we must also, I think, acknowledge that there exists a level of advancement that is worth its sacrifice; and we must hold the AI companies to that level of advancement. We cannot let them distract us with lofty, world-changing promises and then have them do nothing but destroy our culture. We cannot be promised utopia and be delivered slop.


Molly White is a passionate and prolific contributor to Wikipedia, one of the greatest success stories of both the open web and humanity more broadly. She’s also a prominent critic of the Web3/crypto movement and its corruption and financialisation of these open principles.

Molly suggests a series of moments at which someone who was in favour of open access might find that the very openness they support has enabled some questionable business models that violate the spirit of openness:

“When a passionate Wikipedian discovers their carefully researched article has been packaged into an e-book and sold on Amazon for someone else’s profit? Wait, no, not like that.

“When a nature photographer discovers their freely licensed wildlife photo was used in an NFT collection minted on an environmentally destructive blockchain? Wait, no, not like that.”

What do you do in response? Do you try to take back control, to shut things down? Tighten up licensing? Molly is sceptical of both the practicality and desirability of doing so:

“Frankly, if we want to create a world in which every single human being can freely share in the sum of all knowledge, and where education, culture, and science are equitably shared as a means to benefit humanity, we should stop attempting to erect these walls. If a kid learns that carbon dioxide traps heat in Earth’s atmosphere or how to calculate compound interest thanks to an editor’s work on a Wikipedia article, does it really matter if they learned it via ChatGPT or by asking Siri or from opening a browser and visiting Wikipedia.org?”

She observes the same issues that Robin does about the relationship between AI businesses and the commons, namely that it’s an exploitative one:

“Anyone at an AI company who stops to think for half a second should be able to recognize they have a vampiric relationship with the commons. While they rely on these repositories for their sustenance, their adversarial and disrespectful relationships with creators reduce the incentives for anyone to make their work publicly available going forward (freely licensed or otherwise). They drain resources from maintainers of those common repositories often without any compensation. They reduce the visibility of the original sources, leaving people unaware that they can or should contribute towards maintaining such valuable projects.”

But, Molly argues, it’s in AI companies’ interests for the commons to continue to thrive and to continue to be a source.

So, again, we end up in a similar place. Does the existence of AI create something new, that sits alongside the commons and augments it without harming it? Or does it squeeze out human culture, strangling it and supplanting it?

Molly is realistic. It’s unlikely that AI companies will suddenly develop a conscience, and the temptation to be a free rider will remain enormous. Instead, we need big players in the open world – like Wikipedia – to use their power to impose terms on AI companies, to make what they do helpful rather than harmful to commonality, and to make sure they engage with the commons on the commons’ terms.

So, again, I think this is a useful principle: we must whatever power we have to force those who would use the commons to at the very least do so in a non-destructive way, but ideally also to ensure its viability long into the future. AI companies have to own their externalities and preserve the soil from which their products grow.


Kevin Kelly is the founding editor of Wired magazine and a Silicon Valley guru. He’s also been thinking along these lines, suggesting – perhaps just as a thought experiment – AI as a public utility:

“Imagine 50 years from now a Public Intelligence that was a distributed, open-source, non-commercial artificial intelligence, operated like the internet, and available to the whole world. This public AI would be a federated system, not owned by any one entity, but powered by millions of participants to create an aggregate intelligence beyond what one host could offer. … [It] would be open and permissionless: any AI can join, and its joining would add to the global intelligence. It would be transnational, and wholly global – an AI commons. Like the internet, it would run on protocols that enabled interoperability and standards.”

This communal aspect of AI would, Kevin argues, make us more comfortable with opening up more sources of information for its training. Not just copyrighted cultural works, but all sorts of environmental and scientific data too, so the collective AI could respond to everything from census information to realtime weather data.

This feels, as you’d expect from Kevin, techno-utopian. We’ve seen in the real world that there tends to be a Matthew effect (the rich get richer, the poor get poorer) with new technologies. Systems that begin as distributed, federated open standards tend to centralise around powerful players, like email did with Gmail. Technology doesn’t exist in a vacuum; it’s affected by real-world power dynamics.

So far, AI appears to be following the dystopian rather than the utopian path. The power to create foundational models is for the most part concentrated in a few companies, even if there has been some success with open-source models. The dream of many seems to be a world in which AI can be used to substitute capital for labour, further increasing the returns to wealth.

But there’s something useful in setting up this utopian vision, even if it’s not quite practical: it gives us a point of comparison against which to compare reality and a means by which to critique it. It is desirable for AI to be as distributed and federated as possible, because AI confers power and it’s generally in the democratic interest for power to be diffuse rather than concentrated. And so we should make choices wherever possible that favour decentralisation and diffusion.


I think these principles are useful. We have to hold AI to a high standard, we have to prevent it from destroying the commons on which it builds and has already been built, and we have to ensure its power is as federated and distributed as possible.

In the spirit of Tony Benn’s five questions to the powerful, perhaps we need five questions to ask of new AI products:

  1. Is it worth it?
    What genuinely new and useful things does this enable humanity to do? How much does it improve lives?
  2. Is it consensual?
    How permissively does this consume the commons?
  3. Is it sustainable?
    How does this preserve the commons? How does it impact the planet?
  4. Is it permissive?
    Who controls how and where this is used? Who can modify it and build upon it? Is it distributed or centralised?
  5. Is it fair?
    How does this make money, and for whom? Whose labour does it consume and at what cost?

That leads to the practical question: who is asking these questions at the moment? Who is holding the people involved to account? The UK government has been undertaking a consultation on AI and copyright which has widely been seen as selling out the commons to AI companies by carving out a huge exemption to copyright law for their benefit. If it makes the UK a destination for AI investment, it will be because it wins a race to the bottom on IP rights. It is at best attempting to address some narrow questions of intellectual property rights, but it seems to be missing the bigger picture.

Imagine, then, a more ambitious approach. A government that facilitates the creation of something closer to Kevin’s techno-utopian dream. A government that creates incentives for sustainable and consensual AIs, that secures a positive feedback loop from the value extracted from the commons back into the commons. That feels like something that could make the UK “among the leading AI nations” in a way that could make us feel proud, rather than dirty.