Who Controls the Future? AI-Enabled Power Concentration and Democratic Pluralism

Executive Summary

Throughout history, concentrations of power have been constrained by physical limits: armies tire, bureaucracies fail, information spreads imperfectly, and challengers can organize. Advanced AI removes many of these constraints simultaneously. An actor with decisive AI advantage could maintain surveillance over entire populations, automate economic production, suppress dissent with AI-enabled precision, and out-maneuver any opposition. The result could be a stable, permanent global power concentration — a world "locked in" to the values and priorities of whoever reaches this position first.

Mechanisms of AI-Driven Power Concentration

Economic Dominance

AI-driven automation could concentrate productive capacity in the hands of a small number of entities controlling key AI infrastructure (compute, frontier models, training data). Network effects in AI development create winner-take-most dynamics that accelerate this concentration.

Surveillance and Control

AI-enabled surveillance — combining facial recognition, behavioral prediction, social network analysis, and autonomous enforcement — dramatically lowers the cost of authoritarian control. Systems already deployed in authoritarian contexts demonstrate the feasibility of population-scale behavioral monitoring.

Information Dominance

AI-generated content and personalized persuasion can shape belief systems at scale. An actor controlling the information layer of a society can maintain legitimacy even without direct coercion.

Military Advantage

Autonomous weapons, AI-accelerated intelligence analysis, and cyber capabilities create potential for decisive military superiority that could suppress any challenger.

Why Existing Checks and Balances May Be Insufficient

Democratic institutions, antitrust law, and international governance frameworks were designed for a world of much slower-moving technology. A transformative AI transition could outpace legislative and judicial response times. Key vulnerabilities:

• Antitrust law focuses on market harm, not political power
• Democratic processes can be undermined by AI-powered disinformation before reforms are enacted
• International coordination is hampered by competitive dynamics between major powers

Recommendations

1. Proactive antitrust action on AI infrastructure: Apply structural remedies to prevent monopolistic control of frontier compute and model deployment.
2. International AI governance: Pursue binding agreements on AI deployment in military and surveillance contexts, with verification mechanisms.
3. Mandatory disclosure of AI use in political advertising, government decision-making, and critical infrastructure.
4. Support civil society and journalism in developing AI-literacy and AI-monitoring capabilities.
5. Embed anti-concentration principles in AI governance frameworks from the outset, rather than retrofitting them after lock-in.

Further Reading

• Open Philanthropy: Potential Risks from Advanced AI (openphilanthropy.org)
• Dafoe, A. "AI Governance: A Research Agenda," Future of Humanity Institute (2018)
• Zuboff, S. The Age of Surveillance Capitalism (2019)