# On Bargains, Curses, and Dividends

> Two pieces this spring made me think about the bigger picture: on AI, ownership, and the bargain that built every right we hold.

- URL: https://thirdeyefiles.net/on-bargains-curses-and-dividends
- Author: Pablo Núñez
- Published: 2026-06-13
- Topics: Entanglements

---
Recently, discussions about AI have moved from its capabilities to associated costs like environmental effects, privacy concerns, job loss, media distrust, and influence on democracy. While digital heritage and the creative industries—my areas of expertise—are part of these debates, two pieces I came across this spring brought that wider context into focus. The first is [episode 469 of Sam Harris's *Making Sense* podcast](https://www.samharris.org/podcasts/making-sense-episodes/469-escaping-an-anti-human-future), a long conversation between Sam Harris, the American neuroscientist and philosopher, and Tristan Harris, the technology ethicist who co-founded the [Center for Humane Technology](https://www.humanetech.com/). The second is the historian Rutger Bregman's Substack essay "[An Inconvenient Truth About AI](https://rutgerbregman.substack.com/p/an-inconvenient-truth-about-ai)".

All three authors share a common background: they are liberal, secular, and focus on societal issues. However, they reach the present moment by different roads. Sam and Tristan Harris, through AI safety and the psychology of attention, critique how the danger primarily affects our minds. Bregman, from an economic history perspective, views the danger as stemming from shifts in the distribution of power. The first two authors worry about what the machine does to us, while the latter worries about what its owners do to everyone else.

While reviewing their arguments, I noticed that the two pieces repeatedly invoked the same three words—bargain, curse, and dividend. As I mentioned earlier, public debate and conversations in my field typically centre on issues such as environmental costs, data privacy, and jobs, yet words like bargain, curse, and dividend appear to reflect the tone of a deeper conversation, exposing a social disconnect that might be better resolved through transactional solutions rather than through the injection of ideological or moral values. What caught my attention is that words like bargain, curse, and dividend are part of a repertoire of exchange, and I think that such emphasis is a good thing. Those words, subtly, invite mobilisation rather than settling for the understanding, necessary as it is, that currents of thought such as Explainable AI prioritise. Making a system legible matters, but legibility alone does not prompt action; the language of bargains and dividends does, because it names something that can be claimed, withheld, or renegotiated. That's why, in this article, I traced those words to sharpen what I perceive as a common underlying claim: we should not reject AI; we should own it.

## The bargain no one signed

"AI is the ultimate devil's bargain," Sam Harris says: a positive infinity of benefit thrown at your brain, set against a negative infinity of risk. For him, the asymmetry is the point, because the upside, no matter how good it can be, does not cancel the downside. And the deal keeps sweetening. The better the tools get at things that could build a healthier, more representative society, the less you can refuse them, until you forget that no one signed any contract, and yet no one can drop out of it.

Bregman reaches for a different bargain, the fiscal bargain he sets at the heart of every democracy. Rulers have always needed three things from the people they rule: money, labour, and legitimacy. Good government, in his telling, was never philanthropy but structural dependency. Taxes had to be bargained for, and the right to decide how they were spent came out of the same negotiation; so did the duty to send a son to war and the right to ask whether the war was worth fighting. "No taxation without representation," he writes, "but also no representation without taxation." That exchange built every right we hold. And AI, he argues, dissolves it: when the machines do enough of the work and wealth accumulates among their owners, rulers no longer need the rest of society as workers, soldiers, or taxpayers, and the bargain lapses.

Marx had seen the mechanics of this long before the machines could carry them out. Labour, he argued, is the one source of value available to those who own no means of production, and therefore the only leverage the majority holds. Remove the need for that labour, and the leverage goes with it. There is nothing left to bring to the table.

Even though Marx was by no means a liberal democrat, liberal democracy functions on the same principle he so well described. What's unique to it is what Yuval Harari refers to as democracy's self-correcting mechanisms—like election systems, courts, and a free press (check my post on his book [Nexus](/nexus-more-data-less-wisdom)). Although these mechanisms allow the system to self-regulate and correct its mistakes, they only work because the rulers depend on the governed's legitimisation. In a recent conversation with [Ezra Klein](https://www.nytimes.com/2026/05/26/opinion/ezra-klein-podcast-yuval-noah-harari.html?unlocked_article_code=1.p1A.-uYJ.RjCwcXa8RJ4r&smid=url-share), Harari pushes the thought further: imagine a state like Qatar, rich in money and energy and short of citizens, that looks at AI and sees not a workforce but a population, "millions of AI citizens paying taxes" and building companies, a synthetic public that owes the state nothing and asks for nothing. When rulers no longer need society to legitimise their power, representation does not vanish but remains on the books as a set of empty values left to correct the imbalances of social classes that cannot afford to resist the injustice inflicted upon them.

Two bargains, one word. Sam Harris's is existential and individual: it involves the psychological conflict of a deal whose stakes you can't intellectually resolve. Bregman's is structural and collective: it concerns the political economy of representation, which is fundamentally at risk. The two do not exclude each other; instead, they compound.

>> If machines do enough of the work, governments and corporations will no longer need citizens as workers, soldiers, taxpayers, or voters. The bargain does not weaken; it evaporates.

## The curse, or when owners stop needing us

Both pieces reach for the same source to name AI's curse: a paper by Luke Drago and Rudolf Laine called "[The Intelligence Curse](https://intelligence-curse.ai/)."

This concept originates from the better-known *resource curse*, the pattern by which a country that relies on a single extractable resource for its wealth—rather than the labour and taxes of its citizens—tends to diminish its need for its people. Oil-dependent states exemplify this pattern. Wealth flows from the ground, not from the population, eliminating the structural necessity to invest in education, consultation, or representation. Consequently, the political and fiscal agreement that typically links rulers to their citizens never materialises, since the dependency that would establish such bonds is absent.

Tristan Harris carries the idea over to AI: if machines do enough of the work, governments and corporations will no longer need citizens as workers, soldiers, taxpayers, or voters. The bargain does not weaken; it evaporates. Bregman arrives at the same place through the Gilded Age, the late-nineteenth-century boom that built vast fortunes for a handful of American industrialists and financiers. Wealth, says Bregman, is more concentrated now than it was even then: in 1910, at the height of that era, the richest 0.00001 per cent of American households held a fortune worth four per cent of national income; today that same group holds twelve per cent, three times as much. AI is about to make it worse, because the people who own the machines and the software are a smaller group than any owning class in history.

Both Harris's and Bregman's images depict a similar scene: a small group of soon-to-be trillionaires keeping their wealth, while everyone else becomes redundant—not in a dignified, retirement sense, but facing a tough, anxious, gig-economy-driven future.

## The dividend, and who keeps it

And then, by different paths, both pieces arrive at Norway: the petrostate that breaks the pattern.

Norway sits on North Sea oil, a typical example of the resource curse, but it avoided falling into it. The key difference lies in ownership: instead of allowing private control over the wealth, the Norwegian government took public ownership and directs the profits—through its sovereign wealth fund—back to its citizens. The reliance on resource wealth was not eliminated but reshaped within new conditions. Tristan Harris describes this choice as creating an "intelligence dividend, not a curse."

Bregman approaches the dividend from a labour history perspective. He revisits the economist John Maynard Keynes, who in 1930 predicted that increasing productivity would give future generations a fifteen-hour working week by 2030, with machines handling the rest. Keynes was correct about productivity growth but incorrect about who would benefit from it. By the 1980s, a fifteen-hour working week was technically possible; however, it never materialised because the gains went to capital, shareholders, and a rentier class, while wages stagnated and working hours increased. The benefits of progress were real but not equitably shared. Bregman uses compelling charts to make his point, showing that productivity continues to rise long after wages have stabilised at a level below it.

Both pieces reach Norway for one reason. Whether you arrive through AI safety, as Sam and Tristan Harris do, or through labour history and political economy, as Bregman does, the cornerstone is the same: ownership.

>> The unsigned bargain still includes a second party, which is us—unless we quietly accept being sidelined

## Why ownership is the rational position

Neither piece concludes with despair, which is the aspect worth dwelling on, as they present optimism not as a feeling we need to create but as a position with rational grounding.

Bregman concludes by invoking Oscar Wilde, who in [*The Soul of Man Under Socialism*](https://en.wikipedia.org/wiki/The_Soul_of_Man_Under_Socialism) envisioned a future in which intelligent machines would belong to everyone. He draws on history to show that such progress is achievable: the eight-hour workday was not granted voluntarily but achieved through persistent organising and strikes that forced concessions from industrial owners. These patterns, though they may appear novel, have recurred across time, proving that a shorter working week and a shared technological dividend remain realistic demands—even under harsher conditions and fiercer opposition. As Tristan Harris observes, clarity creates agency: once we recognise where the current road leads, we retain the power to choose another.

This brings me back to the original idea. The unsigned bargain still includes a second party, which is us—unless we quietly accept being sidelined. Ownership is the key to regaining legitimacy, which was never freely granted. Rights such as the eight-hour workday, voting, and sovereign funds were gradually seized by those who refused to accept just governance. The same applies to these machines. We won’t own them by rejecting them or waiting for their owners to share them, but by learning to use them effectively so they become part of our work, memory, and creativity—rather than something done to us. This is the demand worth organising around, and it begins, unglamorously, with learning. From where I sit, inside the arts and humanities, the challenge and opportunity still remain at the same crossroads. Which one we take depends entirely on whose hands are on the tools.