Do AI agents trump human agency? Monasterio Astobiza, Aníbal Artificial intelligence Collective behavior Decision Making Processes This publication forms part of grant AutAI PID PID2022-137953OB-I00 funded by MICIU/AEI/https://doi.org/10.13039/501100011033 and by the European Social Fund; and grant PID2024-156166OA-I00, funded by MICIU/AEI/https://doi.org/10.13039/501100011033 and by the European Social Fund Plus (FSE+). This article has also been funded by the Consejería de Universidad, Investigación e Innovación (GrantEMEC_2023_00442) Artificial agents are examined within simulated environments to elucidate the emergence of collective behaviors and decision-making processes under diverse environmental pressures and population structures. Using an Agent-Based Modeling (ABM) framework, the simulation tracked 157,097 iterations across four agent types: cooperators, defectors, super-reciprocators, and free riders; while analyzing 12 core metrics, including alignment indices, coherence, and environmental stress. The results revealed distinct phase transitions in behavior, with low-density populations (d < 0.4) supporting strong consensus formation and higher densities (d > 0.8) leading to fragmentation and increased competition. Agent alignment consistently ranged between 0.28 and 0.37, reflecting partial but stable consensus across conditions. Cooperative behaviors emerged and persisted only when resource availability exceeded a critical threshold (RG ≥ 6), underscoring the role of resource abundance in sustaining collective intelligence. Through interventions such as network topology changes and cognitive plasticity adjustments, agents demonstrated emergent behavioral patterns that arose from their rule-based interactions. It is important to note that these patterns, while complex, do not constitute “sophisticated decision-making” in the sense of genuine intelligence or understanding. Rather, they represent emergent properties of the system, collective behaviors that cannot be reduced to individual agent rules but emerge from their interactions under specific environmental conditions. This distinction is crucial for avoiding misattribution of intelligence to rule-following systems, even when those systems produce complex outputs. These findings provide insights into the mechanisms driving emergent intelligence in artificial systems and their implications for the governance and ethical design of future AI agents. 2025-12-03T09:34:23Z 2025-12-03T09:34:23Z 2025-11-21 journal article Astobiza, A.M. Do AI agents trump human agency?. Discov Artif Intell 5, 348 (2025). https://doi.org/10.1007/s44163-025-00608-y 2731-0809 https://hdl.handle.net/10481/108544 10.1007/s44163-025-00608-y eng http://creativecommons.org/licenses/by-nc-nd/4.0/ open access Attribution-NonCommercial-NoDerivatives 4.0 Internacional Springer Nature