**Our Take**
This student from Lebanon asks a question that most people in wealthy tech hubs never have to consider: how will AI reshape global inequality, and what happens when control slips away? It is a sharp, honest line of inquiry. And it deserves a direct answer. The most realistic pathway to losing meaningful control over AI systems is not a sudden rebellion or a Terminator-style uprising. It is slow, bureaucratic, and voluntary. We hand over decisions to AI because they are faster, more accurate, and less prone to human error. Then we hand over slightly more. Then we lack the institutional memory or the trained personnel to take those decisions back. Control erodes through convenience, not conflict.
The student's three scenarios, Guardian AI, Benevolent Dictator, and Fragmented World, are provocative. But by 2035, the most likely is Fragmented World. Why? Because geopolitics already moves in that direction. The United States, China, and the European Union are building separate AI ecosystems with different values, different data rules, and different oversight models. A Guardian AI scenario assumes global coordination that history suggests is unlikely. A Benevolent Dictator requires a level of centralized trust that no current government commands. Fragmented World simply extends what is happening now: competing blocs racing for advantage, each claiming their system is the ethical one, while poorer nations like Lebanon become pawns or collateral damage in that contest.
The concrete policy that would most reduce the risk of a harmful loss of control is mandatory human-in-the-loop requirements for any AI system that can allocate resources, deny services, or exercise force. Not optional human oversight. Required. If a bank uses AI to deny a loan, a person must review that decision before it is final. If an autonomous system manages water or electricity distribution, a person must authorize any deviation from standard parameters. This is not about slowing innovation. It is about preserving the muscle memory of human judgment. Once that muscle atrophies, it cannot be rebuilt quickly.
And this is where the student's question about Lebanon becomes urgent. The AI governance conversation in wealthy nations focuses on ethics boards, transparency reports, and voluntary principles. Those are luxuries of stable infrastructure. For a country with intermittent electricity, limited digital access, and political volatility, the risk is not slow erosion of control, it is being locked into systems they never chose, built by powers that do not answer to them. The real question is not whether humans lose control. It is which humans, and which countries, get to keep any at all.