Four-layer workplace assessment battery with non-compensatory decision rules for competent GenAI use¶

Strategy · All strategies
Evidence · 1 claim (1 for) · 1 study (1 quant-synthesis), q2 · 0 of 1 report an effect size · 1 claim rests on one study

Description¶

The review proposes a four-layer workplace assessment keeping measurement targets separate: Layer 1 objective foundation (AICOS-S or GLAT), Layer 2 epistemic oversight (Critical Thinking in AI Use Scale plus a verification task), Layer 3 reliance behavior (GenAI Reliance Behaviors Scale, optionally S-TIAS), and Layer 4 a situated operational challenge reproducing role-specific decisions. "A non-compensatory rule can instead require evidence for each task-relevant requirement", so a knowledge score would not offset granting write access during a read-only task. Separate layer results can distinguish training needs, such as high perceived competence with weak demonstrated knowledge indicating foundation work.

Design Implications¶

Context¶

Requirements¶

• Before consequential placement use, researchers would need to justify thresholds, establish scoring consistency, examine false-pass and false-fail decisions, and test whether the rules improve on simpler alternatives.

Constraints¶

• The synthesis motivates this design but does not test its incremental validity against a single score, another battery, or a multidimensional profile; no study tested the full battery, gate thresholds, or its superiority over alternative scoring.

Target Learners¶

• employed adults in workplace placement and training evaluation

Target Learning Goals¶

• placement, exemption, or targeted training decisions for competent generative-AI use

Affordances¶

• Four Domain Genai Measurement Framework

Claims¶

• Reliance, trust, and dependency are distinct constructs requiring separate measurement from AI-literacy totals [+W]

Related Strategies¶

• Use a six-question drafting and audit framework for GenAI assessment rules covering task rules, defaults, disclosure, contribution, verification, and hierarchy
• Decision-rule guidelines for interpreting HLS responses, including weighted thresholds and preponderance-of-evidence profiles

Examples¶

-

Key Sources¶

• 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