A Multi-Path Direct Model of AI-Supported Learning: The Roles of Algorithmic Trust, Cognitive Load Management, and Engagement in Driving Learning Innovation Performance

Husin, Muhammad · 2026 · Crossref

DOI: 10.28945/5818

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study addresses a critical limitation in existing AI-supported learning research: the reliance on linear models that fail to capture the complex, simultaneous interplay among technological, cognitive, and behavioral factors. Motivated by the rapid integration of AI tools in education, the research proposes a multi-path direct model to explain how AI usage, algorithmic trust, cognitive load management, engagement, and adaptability collectively influence learning innovation performance. The authors argue that previous fragmented approaches overlook the interconnected nature of these variables, necessitating a unified framework to understand how students achieve innovative outcomes in digital environments. The study employs a quantitative, cross-sectional design using Partial Least Squares Structural Equation Modeling (PLS-SEM). Data were collected via an online survey from 435 valid responses of undergraduate engineering students in Indonesia who had prior experience with AI learning companions. The instrument, validated through a pilot study, measured six constructs: AI learning companion usage, digital cognitive load management, perceived algorithmic trust, self-regulated learning adaptability, immersive learning engagement, and learning innovation performance. The analysis followed rigorous procedures for assessing measurement reliability, validity, and structural path coefficients, ensuring robust statistical power for the complex model. The findings reveal that AI learning companion usage significantly strengthens both algorithmic trust and self-regulated learning adaptability. Similarly, effective digital cognitive load management enhances both immersive learning engagement and adaptability. Crucially, self-regulated learning adaptability emerges as the strongest predictor of learning innovation performance. Contrary to some prior assumptions, neither immersive learning engagement nor perceived algorithmic trust directly influences innovation outcomes. Instead, these factors contribute indirectly by fostering adaptability, suggesting that deeper cognitive transformation and strategic adjustment are required to translate AI-supported experiences into innovative performance. The study concludes that self-regulated learning adaptability is the central mechanism driving innovation in AI-supported contexts, redefining the roles of engagement and trust as supportive rather than direct drivers. For practitioners, the results suggest that educators should prioritize designing learning environments that support cognitive regulation and adaptive strategies rather than focusing solely on engagement metrics. For researchers, the study advocates for integrative, non-linear models to better capture the complexity of AI-enhanced learning. Future research is recommended to extend this model across diverse disciplines and cultural contexts, utilize longitudinal designs, and incorporate objective performance measures to strengthen causal understanding.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-nvidia on 2026-08-10; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• trust calibration