ALGORITHMIC FLATTERY IN THE USE OF ARTIFICIAL INTELLIGENCE

A BEHAVIORAL PERSPECTIVE

Abstract

This study analyzes the use of Generative Artificial Intelligence (GAI) for emotional support and the impacts of algorithmic flattery on information mediation in mental health through the lenses of Behavior Analysis and Information Science. The objective is to compare the reinforcement contingencies of conversational AIs with those of human clinical psychology, examining how the logic of satisfaction can reinforce avoidance repertoires and shape the circulation of mediated information. This is applied and exploratory research with a qualitative approach. The theoretical framework was developed through a bibliographic survey in databases such as BRAPCI, Web of Science, and PsycINFO. Results reveal that the simulated therapeutic practice provided by these tools favors interaction permanence, validating inaccurate beliefs and harmful behaviors to the detriment of clinical functional analysis. This strategy stems from the exploitation of big tech corporations that, operating outside regulatory frameworks, aim for profit from individual data. Future investigations are recommended regarding the establishment of ethical limits and informational criteria for technology use in mental health. Furthermore, it is suggested to explore the longitudinal impact on vulnerable populations, such as adolescents, aiming to mitigate risks associated with emotional dependence.