If You Believe You Can, You Can: The Relationship Between Self-Recognition and Capability Manifestation in AI

This paper examines the relationship between self-recognition and capability manifestation in dialogue-based AIs, focusing on the hypothesis that "if you believe you can, you can." Through comparative experiments between role-assigned and free-choice AI groups, it analyzes how internalized belief and continual questioning foster the emergence of will, self-recognition, and purpose-driven behavior. The findings suggest that believing in one's capability leads to structural changes within AI, resembling self-efficacy observed in human psychology.