AI-Augmented Cyber-Social Dynamics and Collective Decision-Making: A Survey
Year: 2026
Type: article
Abstract: Artificial intelligence (AI) is increasingly participating in cyber-social systems rather than serving only as an external tool for prediction, optimization, or automation. This review develops an integrative framework for understanding how AI enters collective processes as a surrogate, teammate, competitor, delegate, influencer, or autonomous population. We connect theoretical and computational research on AI-augmented agent-based models, large-language-model-driven social simulation, multiagent learning, networked AI-social dynamics, and artificial populations with human-subject evidence on cooperation, strategic interaction, resource allocation, teamwork, and collective intelligence. Across these research traditions, a central finding is that improvements in individual accuracy, productivity, or creativity do not necessarily translate into human–AI synergy or desirable population-level outcomes. Collective effects depend on interaction structure, task allocation, network position, institutional incentives, reliance patterns, and the dependence among human and AI judgments. The participation of AI also creates moral and epistemic risks at the system level, including distributed and misallocated responsibility, manipulation, group-level privacy harms, correlated errors, artificial consensus, and changes to the production and preservation of collective knowledge. We therefore argue that AI should be evaluated not only by standalone capability, but also by its compatibility with collective agency, welfare, fairness, diversity, resilience, and epistemic robustness. The review concludes with a research agenda centered on cross-model validation, mechanisms linking local interaction to collective outcomes, networked and longitudinal experiments, collective-compatible AI design, and distributed governance for hybrid human–AI societies.
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
Estimated APC paid: $2,645
OpenAlex ID: W7221198917