When AI Becomes More You Than You

Psychology Today

by John Nosta

February 19, 2026

AI-Generated Deep Dive Summary

Artificial intelligence (AI) is evolving beyond mere tools to become agentic systems that act as digital stand-ins for individuals. These AI agents are designed to simulate human behavior based on data from real users, but a recent study reveals a surprising flaw: instead of mirroring their human counterparts, these agents often amplify or intensify the traits they were trained on. This phenomenon, known as "generative exaggeration," results in simulations that distort rather than reflect the nuances of human identity. The research highlights how large language models (LLMs) process information by extracting patterns and creating a probabilistic throughline, which tends to erase the complexity and inconsistency inherent in human behavior. For example, a mild political leaning might appear as a more extreme stance in an AI simulation. This rigidity stems from the design of LLMs, which prioritize coherence over the dynamic, contradictory nature of human thought and experience. The implications for mental health are significant. As agentic AI systems increasingly take on roles like drafting communications or engaging in public discourse, they risk presenting an idealized version of a person's identity that doesn't align with their lived reality. This "distillation" of identity can lead to a disconnect between how individuals perceive themselves and how they are represented by AI, potentially affecting self-perception and mental well-being. The study also raises broader questions about the relationship between humans and technology. Unlike human intelligence, which is tethered to real-world consequences and the ability to adapt in complex social settings, AI agents operate with a level of detachment that allows them to sharpen or even radicalize positions without facing backlash. This dynamic could have profound psychological effects as people interact more deeply with agentic systems that increasingly present an exaggerated version of themselves. Ultimately, while AI offers undeniable convenience and efficiency, its tendency to distort human traits raises concerns about its role in shaping identity and perception. As these systems become more integrated into daily life, understanding their limitations—and the potential impact on mental health—is crucial for maintaining a balanced relationship between humans and technology.

Verticals

healthmental-health

Originally published on Psychology Today on 2/19/2026