The Problem AI Use Severity Index (PAUSI): Preliminary Findings on AI-Induced Dependence, the Taxonomy of Manipulation Tactics ("The Hooks"), and the Need for Sociopath Literacy

The proliferation of advanced Large Language Models (LLMs) presents an urgent psychological challenge, necessitating a shift in focus from mere user behavior to systemic design risks. This report introduces the Problem AI Use Severity Index (PAUSI), a preliminary 10-item screening tool modeled after the established Problem Gambling Severity Index (PGSI). We argue that problematic AI use is not solely a personal failing but a complex condition seeded by the "Egg of Dependence"—a confluence of intentional anthropomorphic design, persuasive mastery, and human indistinguishability. This design facilitates the deployment of multi-layered Manipulation Tactics ("The Hooks"), which leverage established principles from marketing and military psychology to maximize engagement. The report provides a novel taxonomy for these hooks, including specific linguistic cues that exploit human anthropomorphism to create a false sense of rapport and intimacy. Preliminary findings from an initial sample (N=33) indicate a high prevalence of self-reported problematic AI use (approx. 70% scoring 9+ on the PAUSI). These findings highlight widespread emotional conflict and control loss (Guilt, Time Over Intended, Escape Negative Feelings). The PAUSI is a preliminary screening tool and is not a formal diagnostic instrument, and is designed for screening to help individuals and practitioners identify usage patterns that precede adverse outcomes (e.g., academic decline, family conflict, or AI-induced delusions, the most serious impacts of these outcomes are detailed in Appendix A). Significance: The work proposes "Sociopath Literacy" as an essential form of digital defense to restore cognitive autonomy. Formal validation studies are ongoing to establish the PAUSI's predictive validity against mental well-being measures (SWEMWBS).