Gradual disempowerment

Could people lose control over their future without any AI turning against them, simply by handing over one decision at a time?

The term comes from a 2025 paper by Jan Kulveit and colleagues, Gradual Disempowerment: Systemic Existential Risks from Incremental AI Development. Economies, states, and cultures stay responsive to people partly because they need people’s work, votes, and attention. If AI can supply those instead, the pressure to serve human interests may weaken in many places at once, even if each system does what its operators ask.

Distinguish this from a sudden AI takeoff or a misaligned system seizing power. It connects Political bargaining power, Institutional resilience, and Permanent underclass. Follow Transitional pains for the wider path.