Problem AI Use Severity Index

Are you using AI —
or is it using you?

The PAUSI is a forensic screening tool created by Giselle Fuerte of Being Human With AI. Modeled after the clinically-validated Problem Gambling Severity Index (PGSI), it identifies dependency markers and behavioral risks before they escalate into crises.

As AI becomes embedded in professional life, the line between productive tool use and problematic dependency grows increasingly difficult to detect without an objective measure. The PAUSI provides that baseline.

I. Time & Tolerance
▼
Usage exceeding original intent; requiring greater engagement for equivalent satisfaction
II. Emotional Attachment
▼
Reliance on AI for emotional regulation or feeling the system understands one better than people
III. Loss of Control
▼
Inability to reduce usage; concealing the extent of time spent with AI systems
IV. Consequences
▼
Professional friction, social withdrawal, or guilt arising from AI usage patterns
Scoring Thresholds
0
No problem
1–2
Low level
3–8
Moderate
9+
Problem use
Low level of problems0–2
Moderate level of problems3–8
Problem AI use9+
Note: The PAUSI is a preliminary educational screening instrument, not a formal psychiatric diagnostic tool. Findings reflect correlation with psychological distress, not causation. Consult a qualified professional if you have concerns about your usage.

Explore the Data

◉ 66.7% Problem AI Use  ·  n = 96  ·  Mean Score: 12 / 30
Total Assessments
96
Aug. 2025 – Mar. 2026
Mean PAUSI Score
12.0
Out of 30 possible points
Problem AI Use
66.7%
Scored 9 or above

Distribution by Use Category

Click a segment or legend item to expand detail
66.7% Problem, 27.1% Moderate, 6.2% Low.
Problem AI Use
66.7%
64 records · Mean score: 16
Score 9+: 64 respondents show markers consistent with problematic dependency. Scores are elevated across emotional attachment and loss-of-control dimensions. Items 5 and 8 — concealment of AI use and inability to reduce usage — are the strongest indicators.
Moderate Problems
27.1%
26 records · Mean score: 6
Score 3–8: Early warning signs are present, typically in the time and tolerance dimension. Fewer emotional dependency or consequence markers. This cohort represents the highest-value target for early intervention.
Low Level
6.2%
6 records · Mean score: 2
Score 0–2: Minimal indicators of problematic use. This group represents healthy, bounded AI engagement with clear human–tool distinction. Only 6 of 96 respondents reached this threshold.

Mean Score by Dimension Item

Average response per item across all 96 respondents · Hover for item text
Items 1 and 6 show the highest mean responses.

Score Distribution

Count of assessments at each total PAUSI score value · Drag range sliders to zoom
0 – 30
Score distribution.

Audit Your Risk

Answer each item honestly. Your score updates in real time as you respond. Each item is presented after the previous one is answered.

Live Score
Item 1 of 100% complete
0
out of 30
Awaiting response

Findings & Implications

Click any highlighted statistic to reveal its organizational implications. August 2025 – March 2026 · n = 96

Finding 01

AI dependency is not an edge case

Rather than a bell curve centered on moderate use, scores cluster sharply at the high end — problematic AI use is the statistical norm in this dataset, not the exception.

66.7% of respondents scored 9 or above, placing them in the Problem AI Use category with a mean score of 16 / 30. Implications
Organizational Implication
If PAUSI results within an organization revealed a similar distribution, this would signal a structural challenge — one that cannot be addressed through individual coaching alone. Proactive AI governance and enablement frameworks, deployed before dependency takes hold, are the only scalable intervention. This is the evidence base for a structured AI Readiness function.
structural riskdependency markers
Finding 02

Emotional attachment drives severity

Items 3 ("AI understands me better than humans") and 6 ("I use AI to escape negative feelings") are the strongest predictors of overall PAUSI severity — more so than time overuse alone.

Mean 16 The Problem AI Use cohort scores an average of 16 / 30 — nearly double the 9-point threshold, indicating entrenched dependency. Implications
Organizational Implication
Items 8 (inability to reduce use) and 10 (problems in professional relationships) are least likely to score zero in the high-risk cohort — indicating that by the time a user is identified, harm has already materialized. Early detection via periodic forensic assessment is substantially less costly than downstream remediation.
Key Dimension

The emotional dependency vector is the most clinically significant — and the least visible to managers or HR functions without a validated screening instrument.

Finding 03

Moderate users represent the highest intervention ROI

27.1% of respondents scored 3–8 (mean: 6). This cohort sits at a measurable inflection point: warning signs are present, but the dependency has not yet consolidated.

27.1% of respondents fall in the moderate range — the highest-leverage cohort for structured, preventive intervention. Implications
Organizational Implication
Structured usage policies, bounded AI workflows, and role-specific enablement training deployed at this stage can prevent migration into the problem category. The cost of intervention here is a fraction of remediation after dependency has consolidated. This is the operational case for proactive rather than reactive AI readiness programs.
Recommended Model

Position the human as Architectural Lead (sourcing, analyzing, formulating decisions) and AI as Processor (synthesizing, drafting, refining). This enforces cognitive boundaries and prevents the Persona Drift that leads to dependency loops.

Finding 04

Periodic forensic monitoring is operationally necessary

Consistent assessment volume across the seven-month window confirms that dependency risk is persistent rather than situational. One-time assessments are insufficient for institutional health monitoring.

7 months of continuous data enables baseline tracking, longitudinal comparison, and early detection at the team or department level. Implications
Organizational Implication
A quarterly PAUSI cadence, segmented by role and tenure, yields an objective behavioral signal that productivity metrics alone cannot provide. This infrastructure distinguishes organizations that actively govern AI risk from those that merely adopt AI tools. The goal is Cognitive Sovereignty: humans remain the architectural decision-makers, with AI serving as a precision instrument rather than a surrogate judgment system.
quarterly cadencecognitive sovereigntyAI governance