Reliance, trust, and dependency are distinct constructs requiring separate measurement from AI-literacy totals¶

Claim · All claims
Evidence · 1 study (2 entries) · quant-synthesis r1 · q2–q3
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q3 i? The GenAI Reliance Behaviors Scale validated four factors of reliance during problem solving: reflective, cautious, thoughtless, and collaborative use, with overall internal consistency α = .84. → Daniele Verí 2026 q2 i? Trust scores predict intention to rely but are not a competence score, and the Generative AI Dependency Scale describes a maladaptive outcome pattern associated with lower task performance and critical thinking. → Daniele Verí 2026 (2)

Evidence¶

Daniele Verí 2026¶

Daniele Verí. (2026). Beyond AI Literacy: A Structured Review and Exploratory Meta-Analysis of Measures for Competent Generative-AI Use. arXiv. https://arxiv.org/abs/2609.15624 doi:10.48550/arxiv.2609.15624

q3 · i? · quant-synthesis · r1

Scale-validation evidence as reported in the review: exploratory factor analysis used 800 responses, with confirmatory analyses on holdout samples of 730 and 1,173 responses from a second problem-solving activity. The review notes the scale sits closer to competent collaboration than a generic trust score.

"The four factors were reflective, cautious, thoughtless, and collaborative use; overall internal consistency wasα=.84[Hou et al., 2025]."

Daniele Verí 2026 (2)¶

Daniele Verí. (2026). Beyond AI Literacy: A Structured Review and Exploratory Meta-Analysis of Measures for Competent Generative-AI Use. arXiv. https://arxiv.org/abs/2609.15624

q2 · i? · quant-synthesis · r1

Review's synthesis of the Generative AI Dependency Scale (Goh et al., 2025), developed across six studies with 1,333 participants, showing a three-factor structure of cognitive preoccupation, negative consequences, and withdrawal with α = .92 to .93 and test-retest ICC = .87.

"Dependency was associated with lower task performance and critical thinking. It can serve as an outcome or risk indicator and should remain separate from an AI-literacy total."

Discussion¶

Related Claims¶

• Perceived trust mediates the relationship between social influence and generative AI dependence — related
• Epistemic and society-related AI knowledge were negatively associated with trust in GenAI — related
• Nine interaction design factors recur in post-2023 research on trust in AI-enabled systems, with explainability the most studied — related
• Excessive GenAI reliance risks undermining foundational ICT knowledge, supporting supervised in-person assessment of foundational content — related
• GenAI shows a dual-mechanism profile: meta-analytic learning benefits coexist with dependence-related costs to critical thinking — related
• Critical Evaluation and Dependent reliance emerge as empirically distinct constructs, with near-zero correlation between the two facets — related
• Strategic reliance is positively associated with AI literacy, and the reliance types differentiate students across writing process and outcome variables — related
• AI literacy interventions recalibrate reliance on ML systems rather than simply increasing trust — related
• Expectancy-Value beliefs govern the intensity of LLM reliance while AI skills determine its type, indicating the two predictor systems require separate interventions — related
• Strategic users score lowest on standard AI-outcome measures, reflecting instrument validity limits that reward AI-driven outcomes over writing quality — related
• Verification initiation, process quality, and success are distinct constructs that should not be collapsed — a narrower finding that bears on this claim
• Available AI-dependence and attachment instruments follow three non-interchangeable measurement logics and do not converge on one stable construct — related
• The GenAI-RTS is supported by a five-factor structure in which Strategic reliance comprises Deliberate Use and Critical Evaluation facets alongside Instrumental, Dependent, and Dialogic factors — related
• Recurring mechanism clusters — self-regulation, metacognition, critical thinking, self-efficacy, trust, feedback literacy, AI literacy — link GenAI use to agency outcomes — related
• Higher post-task trust in an AI assistant predicts lower appropriate reliance during AI-assisted programming problem-solving — related