Even If the AI Optimists Are Right, It All Might Go Very Wrong
Four futures for post-work America — and our first Riskgaming scenario you can play alone, in your browser, in five minutes
Most writing about AI tries to predict what’s coming next. I want you to decide what’s coming next. So I built Four Forks — our newest Riskgaming scenario.
For this game, we’re pretending the rosiest AI optimists are right. Intelligence becomes even more essential than electricity, rocketing humanity toward a level of productivity and creativity that obviates most human toil.
And we’re further supposing that the AI doomers and existential-riskers are dead wrong. The robots are fully under our control. There are no paperclip maximizers; biosecurity and cybersecurity get locked down tight; misinformation and hallucination and bias — all gone. AI is aligned with us, with no threats to (or dire complications for) humanity, except those of our own making.
But here’s the problem. Even if we take all of this for granted, even if we presume we are in the “good” timeline, that won’t be enough. Because what remains may be the biggest challenge of all for human flourishing: politics.
In the optimistic world, AI generates an economic surplus that frees us from any labor we don’t want to do. But what that abundance feels like varies wildly depending on the answers to two political questions: Who controls the surplus, the state, or private hands? And what does the surplus fund, common civic goods, or individual consumption?
The answers create four forking paths for humanity’s future:
Gulf Rentier: A check arrives every two weeks, and people are taken care of, but live in an atomized society on their phones.
Tech-Feudal: Those who own the models run the show and live like kings. Everyone else subsists in the permanent underclass.
Societal Breakdown: All for themselves, the Hobbesian nightmare.
Civic Abundance: The surplus goes toward public projects; the rich are less well off, but the library is open late, the music festival is free, and the streets and subways (quiet, electric, safe) are fully funded.
You can see my preferred outcome is Civic Abundance, with a vision I’ve previously outlined. Perhaps you prefer a different outcome, and you’d rewrite one of these quadrants with how you see it. What no one wants, though, is for us to sleepwalk into a world we didn’t choose rather than fighting for a future of our own making.
But that’s what’s happening right now. The left is treating AI as either frivolous, fraudulent, or frightening, and refuses to see its opportunity. The right wants to just “let the private sector cook,” consequences be damned. Meanwhile, thousands of hard, unglamorous, and often invisible decisions are deciding our fates.
Four Forks reveals the hard tradeoffs those decision-makers face by putting you in their shoes. When you play, you take one of four seats of power in the near future: the CEO of the second-place AI lab, the host of the most-listened-to AI policy podcast, the president of the United States, or the head of the country’s largest union. You’ve got to balance your own interests against where you’re pushing the world.
Sometimes those incentives align. But as in real life, more often than not they don’t. Does winning mean getting a high score? Or landing the world in an optimal place? That’s up to you.
Every Riskgaming scenario we’ve built before this one needed a room full of players and a facilitator. Four Forks needs only you. It takes about five minutes, in your browser or on your phone: https://fourforks.riskgaming.com/.
Play it, post your results card, ask a friend what they got. Oh, and there are a few secret endings. Good luck finding them.
I really like this short piece, especially because it asks one of the most important questions: if the technology succeeds, what kind of society does the surplus produce? This is exactly the kind of long-term thinking we need, because only then can we ask what is actually best for society and for individuals.
One clarification I’d be interested in is whether the key axis is really “state vs. private hands,” or whether that frame understates the range of institutions that could mediate AI abundance. The non-state side may be more than just model owners or billionaires. It could include corporations, foundations, universities, unions, co-ops, nonprofits, religious institutions, and eventually network states or other sovereign-seeking communities. I’m not saying all of these institutions will survive in the long run, but in the short run, many may matter, especially because I don’t think all jobs will be replaced without major advances in robotics, and some may not disappear even with significant advances. These institutions could produce very different versions of “private” abundance.
I’d also separate control of the AI surplus from control of pre-AI chokepoints. Even if the existing rich don’t own the frontier models, they may still own land, housing, energy, finance, logistics, media, universities, healthcare, political access, and status networks. If intelligence becomes cheap while safety, land, compute, energy, status, human attention, and institutional access remain scarce, old wealth may still capture much of the upside. I don’t think AI abundance will erase old wealth for several generations in many cases; it may even make old chokepoints more valuable in the short term.
We also do not yet know which layer will capture most of the gains: model owners, compute and infrastructure providers, companies building on top of models, incumbent firms applying AI in fields like drug discovery or finance, or some new type of institution that does not yet exist.
Another related issue is exit power. Again, I may have misunderstood the post, but the frame seems to assume that the surplus can be politically allocated by either the state or private hands. But ultra-rich AI actors may not simply remain governable by existing city-states. They may jurisdiction-shop, move IP and compute, bargain for special zones, build network states, or use the threat of exit to limit taxation and regulation. So the question shifts from only who controls the surplus to also whether surplus-holders can escape democratic claims on it.
Network states are especially interesting because they blur the categories. I’m not advocating for them, and I don’t personally plan to join one even if I could. But in the best case, they might become laboratories of civic abundance: high-trust communities pooling surplus into education, healthcare, housing, research, culture, and shared infrastructure. In the worst case, they become gated micro-sovereignties: abundance for members, exclusion for outsiders, and governance captured by founders, token-holders, or ideological elites.
So the deeper question is: what agency do ordinary people have in the post-work settlement? Are they citizens, co-owners, members, and participants with real voice? Or are they consumers, clients, users, and subjects being managed by whichever institution controls the relevant bottleneck?
I agree that Civic Abundance seems like the right target, but it will likely require more than redistribution alone. It will require institutions that give people voice, ownership, exit, and dignity, as well as democratic states capable of binding mobile capital and quasi-sovereign actors to a shared civic settlement. Otherwise, even a materially generous AI future can still feel like dependency or managed comfort.
And at the individual level, there is another layer: purpose, identity, and meaning. Work currently supplies a lot of that for many people. If AI abundance reduces the need for labor, then the political question of surplus distribution becomes inseparable from the human question of what people do, belong to, and become.