Researcher proposes AI democratization to preserve liberties

In an August 2026 essay titled 'On Democratizing ASI to Preserve Civil Liberties', researcher Michael Dickens argues that unchecked artificial superintelligence (ASI) could undermine liberal democracy by creating insurmountable power imbalances. While conceding that democratizing AI is risky, he proposes a gradualist approach: first distributing advanced AI assistants widely, then incrementally improving capabilities while ensuring no single entity gains decisive advantage.

The core mechanism relies on preventing recursive self-improvement monopolies. By making each generation of AI marginally better than the last, the system creates a stalemate where defecting developers would face coordinated resistance from users with superior previous-generation tools. Dickens acknowledges implementation challenges and ultimately endorses international treaties to pause dangerous AI development, framing democratization as a 'less bad' alternative to unchecked ASI dominance.