The Democratic Republic of Congo sits on an estimated 24 trillion dollars in untapped mineral wealth. By any normal accounting, that should make its people among the richest on earth. Instead, roughly three out of every four Congolese live on less than 2.15 dollars a day, one of the worst poverty rates anywhere.
Economists have a name for this pattern. They call it the resource curse. Countries with abundant natural wealth often do worse for their citizens than countries with none, because the money never had to come from the people in the first place. A government that gets rich from oil in the ground has no real reason to build schools, courts, or roads for a population it does not need to tax.
Luke Drago and Rudolf Laine, in a widely discussed essay series called The Intelligence Curse, argue that advanced AI is about to put much of humanity in the same position. Not as a metaphor. As a mechanism.
Take a population's usefulness away, and the investment in that population stops making sense, whether or not anyone intends it to.
Why do powerful actors invest in ordinary people at all?
Their case is simple and uncomfortable. Powerful actors, meaning states and large companies, do not invest in ordinary people out of goodwill. They invest because people are useful, as taxpayers, as voters, as workers worth competing for. Build good schools and you get a more productive, more taxable population. Pay engineers well and you keep the ones who could otherwise leave for a competitor.
The same logic explains the free meals, the stock options, the company laundry service at a Silicon Valley firm. It is not generosity. It is the cost of competing for people whose skills the company genuinely cannot replace yet.
How completely could AI remove that leverage?
Drago and Laine argue that AGI removes it more completely than any resource before it. Oil runs out eventually, which is part of why petrostates still have some long term reason to diversify. Intelligence does not run out. It keeps improving. A state or company that can generate revenue and labor from AI systems has less and less reason, year over year, to fund the schools, salaries, and safety nets that once made ordinary people worth investing in.
Not every resource-rich country ends up like the Congo. Norway is the case Drago and Laine keep returning to. It sits on enormous oil wealth and used it to build one of the highest standards of living on the planet, with a sovereign wealth fund worth over 300,000 dollars per citizen. The difference was not the resource. It was that Norway already had strong, accountable democratic institutions before it struck oil.
Drago and Laine do not treat the curse as inevitable. They propose averting the worst AI catastrophes without concentrating more control in fewer hands, diffusing AI broadly enough that ordinary people actually benefit from what it can do, and democratizing the institutions that will decide how the transition gets managed, before those institutions stop needing anyone's vote to function.
That comparison is the uncomfortable part of the argument. Good institutions can blunt a resource curse, but few countries have Norway's starting conditions, and AGI will not wait for the rest of the world to build them first. Worse, because intelligence never runs dry the way oil does, even the long-run pressure that eventually pushes petrostates toward economic diversification will not apply here. There is no peak-AI moment forcing a rethink.
We have written before about how institutional accountability breaks down once the people running an institution no longer need the people inside it. Drago and Laine are describing the same mechanism at civilizational scale, applied to entire populations instead of individual researchers.
Does the Future Human Hypothesis quietly bet against this?
This is where Future Human Hypothesis has its own stake in the argument, even though Drago and Laine never mention it. Our working premise is that some future version of humanity survives long enough, and coherently enough, to potentially study its own past. That premise quietly assumes something Drago and Laine would recognize immediately: a lineage with enough institutional continuity, and enough ordinary people still meaningfully inside the system, to still be recognizably us, doing something recognizably human, however far downstream.
If the intelligence curse runs all the way to its logical endpoint, permanent concentration of power in a handful of AI-saturated actors, most of humanity priced out of any real bargaining position, that is not obviously a lineage capable of becoming careful, patient observers of its own deep past. A civilization that stopped needing most of its own people may not be a civilization with much continuity left to hand down.
None of this is a prediction that the curse will run its full course. Drago and Laine are explicit that their case is a warning about a mechanism, not a forecast of an outcome.
But the shape of the bet is worth naming plainly. If future humans, in whatever form the hypothesis imagines, turn out to be real, that is a small, strange piece of evidence that the curse got broken somewhere between here and there. If they are not, the reason may not arrive as a single dramatic event. It may simply be a long, well-documented loss of any reason for anyone powerful to keep us around.