Free AI Dependency Test (GAID)

Check in on your relationship with ChatGPT, Claude, Replika or other generative AI. Free, anonymous, 2-3 minutes.

The GAID — Generative AI Dependency Scale, scientifically validated (Goh et al., 2025)

Start the test now

GAID — Generative AI Dependency Scale

GAID is an 11-item self-report measure of generative AI dependency, validated in 2025. It captures the core features of behavioural dependency adapted to generative AI use — cognitive preoccupation, negative consequences and withdrawal — across tools like ChatGPT, Claude, Replika and DALL-E. Scores from 1.00-5.00 fall into three bands: low, moderate, high.

  • 11 scientifically validated items across 6 studies, n=1,333 (US + Singapore)
  • Takes about 2-3 minutes to complete
  • 100% free and anonymous — no email or account required
  • Three dimensions: cognitive preoccupation, negative consequences, withdrawal
  • Excellent internal reliability: α = .92–.93, ICC = .87
  • AI overuse often masks anxiety, avoidance or low self-confidence
Goh, A., Hartanto, A., & Majeed, N. (2025). Development and validation of the Generative AI Dependency Scale (GAID). Computers in Human Behavior Reports.

What the GAID test measures

Three psychological dimensions that define behavioural dependency on generative AI.

Cognitive preoccupation

Persistent mental engagement with AI: anticipation, the urge to use it, growing influence on everyday decisions.

Negative consequences

Self-reported concerns about dips in performance and confidence in your own abilities.

Withdrawal

Psychological and affective discomfort when access to generative AI is restricted.

Signs of problematic use

If several of these patterns sound familiar, GAID can help you understand your relationship with AI more clearly.

AI is your first reflex

For any non-trivial task, your first move is to open ChatGPT or Claude.

You feel less capable without AI

You feel you can't complete work or make good decisions without an AI model's help.

Discomfort without access

You feel restless, agitated or irritable when access to AI is restricted (offline, no signal, subscription lapsed).

Loss of your own skills

You've noticed your problem-solving or writing skills have dropped since you started using generative AI heavily.

Use outside the task

You open AI even for tasks where it isn't needed or that have no connection to the work in front of you.

Others around you have noticed

Family or colleagues tell you that you seem to rely on AI too heavily or seem constantly preoccupied with it.

Ready to check in on your AI use?

11 statements, 1-5 scale, instant results with a three-dimension breakdown. Free, anonymous, no account needed.

Start the GAID test

Sample items from the GAID

Cognitive preoccupation

My decisions are often influenced by generative AI.

Negative consequences

I have difficulty completing work or other responsibilities without generative AI.

Withdrawal

I feel restless when I can't use generative AI.

Negative consequences

I feel less confident in my own abilities without generative AI.

About the GAID scale

GAID was developed and validated by Adalia Goh, Andree Hartanto and Nadyanna Majeed (Singapore Management University), published in Computers in Human Behavior Reports in 2025. The scale was tested across six studies with 1,333 participants in the United States and Singapore.

The authors started from an initial pool of 30 items and refined the scale through exploratory and confirmatory factor analysis down to 11 final items, organised across three dimensions: cognitive preoccupation (3 items), negative consequences (4 items) and withdrawal (4 items). The scale demonstrated excellent internal reliability (Cronbach's α = .92–.93) and good test-retest reliability (ICC = .87).

Unlike classic dependency scales (substances, phone), GAID deliberately excluded the ‘tolerance’ and ‘mood modification’ dimensions. The reason: AI becomes more effective over time, so the ‘use more to feel the same’ pattern doesn't apply. Generative AI dependency is understood instead as persistent cognitive preoccupation, functional consequences, and psychological discomfort when access is interrupted.

Frequently asked questions

What is the GAID test?+
GAID (Generative AI Dependency Scale) is an 11-item questionnaire developed and validated by Goh, Hartanto and Majeed (2025) for assessing dependency on generative AI (ChatGPT, Claude, Replika, DALL-E, etc.). It measures three dimensions: cognitive preoccupation, negative consequences and withdrawal. It was validated across 6 studies with 1,333 participants in the US and Singapore, with excellent internal reliability (α = .92–.93).
Is AI dependency a real disorder?+
Generative AI dependency is not officially classified as a disorder in the DSM-5 or ICD-11 (the scale was validated in 2025, so this is a new research category). However, the authors situate it within the behavioural-addictions model (Goodman, 1990) and report significant correlations with anxiety, lower self-confidence and weaker cognitive performance. As AI becomes embedded in daily life, it's useful to understand your own usage pattern.
What does a high GAID score mean?+
A high score (mean above 3.67 out of 5) indicates consistently present signs of dependency across several dimensions: heavy reliance on AI for decisions, discomfort when access is unavailable, or a noticeable drop in confidence in your own abilities. This is NOT a diagnosis, but a signal worth exploring with a therapist or your GP.
Is GAID the same as a smartphone addiction test?+
No. The SAS-SV measures problematic smartphone use in general (compulsive checking, social media, etc.). GAID is specific to generative AI — how you interact with models like ChatGPT, Claude or Replika. The two can coexist, but they measure different behaviours.
How does AI dependency differ from classic dependency?+
The GAID authors deliberately excluded the 'tolerance' and 'mood modification' dimensions. The reason: unlike substances or other technologies, AI becomes more effective over time — the user gets better results with less effort, so the classic 'more to feel the same' pattern doesn't apply. AI dependency is defined more by cognitive preoccupation, functional consequences and psychological withdrawal.
Are there clinical cut-offs for GAID?+
No. GAID is a new scale, validated in 2025. Unlike well-established tests like PHQ-9 or GAD-7, there are no officially recognised clinical cut-offs yet. MatchyMatch uses a low/moderate/high band based on thirds of the 1-5 score range. Treat the result as a self-reflection signal, not as a diagnosis.
What can I do if I have a high score?+
Practical strategies: set AI-free windows (e.g. the first hour of the morning), try to handle simple tasks unassisted before reaching for AI, notice whether your use is masking anxiety or avoidance. If you notice associated emotional discomfort or a drop in self-confidence, a therapist can help you understand which psychological needs are being met through AI and how to build a more balanced relationship. You can also speak to your GP or self-refer to NHS Talking Therapies.
Can AI completely replace a therapist?+
No. Generative AI models can offer information, structure and a space for self-reflection, but they cannot offer the therapeutic alliance, real human presence and clinical nuance that a therapist brings. GAID research also shows that heavy reliance on AI correlates with reduced confidence and wellbeing. Use AI as a complementary tool, not a substitute for human relationships or therapy.

How can a therapist help?

If your score points to moderate or high dependency, a therapist can help you understand which psychological needs are being met through AI and develop strategies for a more balanced relationship with technology. On MatchyMatch you can find HCPC / BACP / BPS / UKCP registered therapists experienced with compulsive behaviours and digital wellbeing.

Find a therapist