"Smarter but Not Wiser": A Transdisciplinary CFA-GRM Simulation Pipeline for Mapping Human–GenAI Interaction

Background The integration of Generative AI (GenAI) in higher education creates a paradox between technical efficiency and cognitive depth. Current institutional responses rely more on fragmented oversight than on ecology- and process-oriented assessment. There is a lack of diagnostic tools to map psychometric thresholds when strategic cognitive shifts transition into functional dependence. Methods We present a simulation-based methodological framework using the Tripartite Confirmatory Factor Analysis–Graded Response Model (CFA-GRM). Grounded in the cognitive-affective-conative tripartite model, this study uses a Monte Carlo simulation (N = 1,000) to demonstrate how latent interaction pathways can be structurally validated and how item parameters can be calibrated against a pre-specified diagnostic threshold ( θ ≈ 0.5), with a replication study to establish parameter recovery. Results Across 200 replications at each of N = 1,000 and N = 300, all 400 converged with admissible solutions and recovered the generating parameters with maximum absolute relative bias of 0.040; interval coverage met conventional criteria for 11 of 15 parameter groups at N = 1,000 and 8 of 15 at N = 300. In the single illustrative dataset, the fit is near-perfect (robust CFI = 0.995, RMSEA = 0.010) because the generating and estimated models coincide, a property of the design rather than evidence for the model. Test information peaks at θ ≈ 0.07–0.22, not at the proposed threshold of θ ≈ 0.5. Furthermore, we introduce a Transdisciplinary Calibration Matrix that translates these psychometric signals into context-specific pedagogical interventions across four epistemic layers, in line with SDG 4. Conclusions This study does not claim empirical generalization. It provides a methodological blueprint, a recovered parameter set, and a shared diagnostic language for future empirical work on the ethical and adaptive integration of GenAI in higher education, aligned with SDG 4.

Authors

Institutions

Publication Details

Journal
F1000Research
Published
2026-10-08
DOI
https://doi.org/10.12688/f1000research.190208.1
Primary Topic
Psychometric Methodologies and Testing
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

"Smarter but Not Wiser": A Transdisciplinary CFA-GRM Simulation Pipeline for Mapping Human–GenAI Interaction

Ifan Rivaldo, Syarifah Niken Purnani, Kana Hidayati, Nur Sehang Thamrin et al.
F1000Research
Psychometric Methodologies and Testing
article

"Smarter but Not Wiser": A Transdisciplinary CFA-GRM Simulation Pipeline for Mapping Human–GenAI Interaction

Ifan Rivaldo, Syarifah Niken Purnani, Kana Hidayati, Nur Sehang Thamrin, Tri Susanti, Putri Theresia Oktaviani Manorek, Barquna Tri Raraswati, Juverio Pangestu, Achmad Fatahillah, Chalimatus Sa'diyah, Rina Nopita Manullang, Jacinto Soares da Silva, Amirudin Amirudin
article en

Abstract

Background The integration of Generative AI (GenAI) in higher education creates a paradox between technical efficiency and cognitive depth. Current institutional responses rely more on fragmented oversight than on ecology- and process-oriented assessment. There is a lack of diagnostic tools to map psychometric thresholds when strategic cognitive shifts transition into functional dependence. Methods We present a simulation-based methodological framework using the Tripartite Confirmatory Factor Analysis–Graded Response Model (CFA-GRM). Grounded in the cognitive-affective-conative tripartite model, this study uses a Monte Carlo simulation (N = 1,000) to demonstrate how latent interaction pathways can be structurally validated and how item parameters can be calibrated against a pre-specified diagnostic threshold ( θ ≈ 0.5), with a replication study to establish parameter recovery. Results Across 200 replications at each of N = 1,000 and N = 300, all 400 converged with admissible solutions and recovered the generating parameters with maximum absolute relative bias of 0.040; interval coverage met conventional criteria for 11 of 15 parameter groups at N = 1,000 and 8 of 15 at N = 300. In the single illustrative dataset, the fit is near-perfect (robust CFI = 0.995, RMSEA = 0.010) because the generating and estimated models coincide, a property of the design rather than evidence for the model. Test information peaks at θ ≈ 0.07–0.22, not at the proposed threshold of θ ≈ 0.5. Furthermore, we introduce a Transdisciplinary Calibration Matrix that translates these psychometric signals into context-specific pedagogical interventions across four epistemic layers, in line with SDG 4. Conclusions This study does not claim empirical generalization. It provides a methodological blueprint, a recovered parameter set, and a shared diagnostic language for future empirical work on the ethical and adaptive integration of GenAI in higher education, aligned with SDG 4.

F1000ResearchVol. 15
Yogyakarta State University (ID), Tadulako University (ID), Universitas Gadjah Mada (ID), Sepuluh Nopember Institute of Technology (ID), State University of Padang (ID), Universitas Sumatera Utara (ID), University of Bangka Belitung (ID), Badan Pusat Statistik (ID)
Openalex Percentile: Top 9%
Psychometric Methodologies and Testing
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.