Of the people, by the algorithm: how AI transforms the role of democratic representatives?

Abstract As artificial intelligence becomes embedded in democratic governance, a fundamental question emerges: how does AI transform the role of political representatives? This review analyzes AI’s impact across two channels: input representation (aggregating citizen preferences) and output representation (implementing policy decisions). It employs five democratic criteria to evaluate impacts, and examines the case studies of Taiwan’s vTaiwan platform and Austria’s AMS algorithmic profiling system. The analysis reveals AI transforms representatives’ roles along both channels: from interpreters of obscure public will to facilitators who reconcile clearly expressed preferences with practical constraints (input side), and from direct decision-makers to architects of algorithmic decision-making (ADM) systems (output side). Six institutional conditions determining whether AI enhances or undermines representation are derived: explicit democratic authorization of objectives, transparency extending to the system design stage, accountability mechanisms enabling challenge of system premises by operators, platform independence with institutional integration, active reduction of participation barriers, and clear authority frameworks preventing selective implementation of citizen consensus.