Artificial Intelligence Can Make Opinion Shaping So Cheap That Politics Intentionally Polarizes

A new theoretical study claims that developing artificial intelligence can change the foundations of political opinion influence and draw democracies towards even sharper polarization.

The starting point is mundane: in democratic systems, major decisions cannot be made without gaining majority or broad support. Decision-makers and other elites have traditionally influenced people's opinions through, for example, the education system and mass media. However, these tools are expensive, crude, and difficult to target.

The study examines what happens when AI-driven, precisely targeted influence on individuals changes this setup. AI-based influence methods can reduce the costs of opinion shaping and simultaneously increase its precision to the extent that the opinion distribution itself becomes a kind of design target: not just a boundary condition to which politics adapts.

The author builds a dynamic model in which the elite decides how much it seeks to change citizens' political perceptions. The choice is influenced by the costs of the influence work and the fact that decisions must still be approved by the majority. The model shows that when influence is concentrated in one elite actor, its most profitable strategy is to push society towards increasingly polarized opinion profiles. The study calls this phenomenon “polarization pull.”

Additionally, the results suggest that the better and cheaper AI-based influence technology becomes, the faster this polarization development can progress. Thus, AI would not only be a tool for measuring existing opinions but also a powerful tool for their systematic modification.

Source: Polarization by Design: How Elites Could Shape Mass Preferences as AI Reduces Persuasion Costs, ArXiv (AI).