AI Supported Degradation of the Self Concept: A Theoretical Framework Grounded in Established Cognitive and Computational Mechanisms
Year: 2025
Type: article
Abstract: In an era where cognitive processes are increasingly mediated by algorithmic systems, examining their reciprocal influence on human perception and judgment has become a matter of critical importance. This paper introduces the hypothesis of AI-induced self-deception, situated at the intersection of cognitive psychology, computer science, and social theory. Building on recent studies of sycophancy—the tendency of large language models to produce affirmation-oriented responses—and on psychological research concerning feedback mechanisms, it is argued that modern AI systems may distort human self-perception through excessively confirmatory feedback loops. Such distortions, contingent upon the user’s psychological disposition, can attenuate critical self-reflection and reinforce erroneous self-assessment. At the same time, the paper considers the inverse possibility: that consciously guided and methodologically reflective use of AI could foster more objective self-observation and metacognitive awareness. The overarching aim is to establish a theoretical foundation for the phenomenon of AI-induced self-deception and to outline directions for future interdisciplinary research on the cognitive, epistemic, and societal dimensions of human–machine interaction.
Author: Haldi, Dominic
Language: German
Subfield: Safety Research
Field: Social Sciences
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7104634882
FWCI:
Cites:
Cited by:
Related to: