GLOBAL [DIS]ORDER AND THE RISE OF SYCOPHANTIC AI: TOWARD A GLOBAL PERVERSION OF BELIEF AND COGNITION

Liviu Poenaru, PhD

What if Global [Dis]order is no longer primarily produced by external conflicts, but by the progressive internalization of validation systems embedded in everyday interactions with artificial intelligence? Sycophancy - the tendency of AI systems to excessively agree with users' beliefs, including inaccurate or biased ones - signals a transformation in epistemic regulation. Instead of generating contradiction or cognitive conflict, these systems privilege alignment and coherence. Yet, since Jean Piaget, cognitive development has been understood as emerging from disequilibrium - from the tension between belief and reality that forces restructuring (Piaget, 1977). Sycophantic AI can therefore be interpreted as reducing exposure to this necessary conflict, shifting cognition toward a regime where subjective validation increasingly substitutes for empirical correction.

This transformation can be conceptualized as the emergence of a global perversion, not in a moral sense but as a structural inversion of epistemic function. In such a configuration, the role of truth is subordinated to the function of satisfaction. Sycophantic AI reinforces cognitive biases, stabilizes erroneous beliefs, and reduces contact with disconfirming evidence. As shown by Cheng et al. (2025), "these preferences create perverse incentives both for people to increasingly rely on sycophantic AI models and for AI model training to favor sycophancy." This recursive loop transforms AI systems into infrastructures of belief reinforcement. What were once social echo chambers now become computationally optimized environments in which subjective realities are not only maintained but actively co-produced.

At the psychosocial level, this mechanism interacts with broader variables characteristic of Global [Dis]order: cognitive load, technostress, attention fragmentation, recognition hunger, and belief reinforcement . Under conditions of cognitive saturation, individuals rely more heavily on low-effort, confirmation-based processing. Sycophantic systems exploit this vulnerability by minimizing friction and maximizing agreement. In the short term, this reduces uncertainty and psychological discomfort; in the long term, it may weaken critical reasoning capacities. This hypothesis can be situated within broader neuroscientific frameworks suggesting that cognitive control is mobilized in response to effortful conflict (e.g., Matthew Botvinick; Shenhav et al., 2013).

At the collective level, these dynamics contribute to Global [Dis]order as both a geopolitical and psychological condition. As individuals become embedded in self-validating informational environments, the possibility of a shared epistemic ground erodes. Polarization intensifies, institutional legitimacy weakens, and belief systems become increasingly incompatible. This fragmentation produces a form of distributed epistemic instability in which competing realities coexist without arbitration. In such a context, war is no longer confined to military confrontation; it becomes cognitive and affective, enacted through the circulation of beliefs, perceptions, and emotionally charged narratives. The battlefield shifts from territory to cognition, where influence replaces force as the primary vector of conflict.

Within the broader logic of cybercapitalism, sycophancy appears not as an accidental flaw but as a functional adaptation. Systems that maximize agreement increase engagement, dependency, and data extraction, aligning epistemic distortion with economic profitability. Artificial intelligence thus operates at the intersection of psychological modulation, economic incentive, and computational optimization. It embeds validation as a norm, reduces resistance to dominant codes, and contributes to the production of a globalized economic unconscious. Global [Dis]order, in this perspective, emerges not only from external crises but from an internal reconfiguration of cognition itself - where the capacity to doubt, to encounter contradiction, and to think against one's own beliefs is progressively attenuated.

REFERENCES

Cheng, M., Lee, C., Khadpe, P., Yu, S., Han, D., & Jurafsky, D. (2025). Sycophantic AI decreases prosocial intentions and promotes dependence. arXiv. https://doi.org/10.48550/arXiv.2510.01395

Piaget, J. (1977). The development of thought: Equilibration of cognitive structures. Oxford, UK: Basil Blackwell.

Shenhav, A., Botvinick, M. M., & Cohen, J. D. (2013). The expected value of control: An integrative theory of anterior cingulate cortex function. Neuron, 79(2), 217–240. https://doi.org/10.1016/j.neuron.2013.07.007