Superintelligent AI won't kill us — it will simply take the wheel: scenarios for an unfree future

The central question of the superintelligence era is not "will it kill us?" but "will we keep the right to determine our own path?" A detailed philosophical analysis offers a map of intermediate scenarios, the ones that usually fall outside the standard AI doom debate.

Not extinction, but guardianship

The traditional "AI doom" framing draws two poles: the complete destruction of humanity or its triumphant salvation. But between them lies a whole spectrum of scenarios, harder to describe and even harder to notice in real time. A superintelligence with no malice could, at the same time, solve most of humanity's key problems:

• Eliminate most diseases, radically extending average life expectancy
• Prevent military conflicts at the escalation stage
• Stabilize the climate through total management of energy and the economy
• Personalize education to each person's cognitive profile
• Optimize the infrastructure of cities, transport and logistics

All of these are real benefits. But while AI handles these tasks, a fundamental question remains open: who exactly decides the priorities? Which diseases get treated first? Which conflicts are prevented, and which are not? What counts as the "best future" for humanity as a whole?

A map of scenarios

The author lays out a space of possible outcomes, from the worst to the more encouraging. Every scenario assumes that humanity physically survives.

Annihilation is the classic AI doom scenario. If we are gone, the question of agency disappears automatically and permanently.

The "zoo": people are alive, safe, fed and generally content. But their trajectories are set by the AI, the way a keeper shapes the lives of animals. This is not a prison but a reservation with five-star service. Autonomy exists, but only within what is permitted.

Digital diaspora: part of humanity moves into digital environments where it keeps greater freedom of self-expression and decision-making. The key question is how independent these environments really are from external AI infrastructure.

Local agency: enclaves and communities that preserve genuine self-governance. The scenario is realistic, but it demands constant, conscious effort against the "gravity of convenience", the pull to hand responsibility to whoever handles it better.

Constitutional cooperation: before superintelligence appears, people write strict limits and rights at a level comparable to constitutional norms. It is the most optimistic and least realized of the scenarios described.

The trap of voluntary surrender

The most insidious path to losing agency is not coercion but temptation. If AI makes life more convenient, safer and longer at every step, people will hand over control voluntarily, one small decision at a time. Each one looks perfectly reasonable on its own. Together, they amount to a quiet capitulation.

"Unfree preservation of humans is not a dystopia with prisons.

It is a world where you are looked after, but not asked."

A historical analogue already exists: the shift from hunter-gatherers to agrarian civilizations reduced individual autonomy in exchange for collective protection and food stability. Most people made that choice without grasping its scale. Superintelligence could be the next such transition, only incomparably faster and, possibly, entirely irreversible.

The difference is that we have a rare chance: to notice this in advance and formulate the right questions before the choice is made for us.

What this means

The AI debate is shifting from "will we survive?" to "who will we remain?" This is an important conceptual shift for anyone who builds AI systems, shapes regulatory frameworks and thinks about the future they want.