Perceived Workload Across Traditional, Gamified and Artificial Intelligence-Supported Learning Conditions: A NASA-TLX Study in Higher Education

Synthesis: Speranza (2026) compares perceived workload across three conditions in a university experiment — traditional individual study, conventional gamification, and gamification with conditioned progression supported by artificial intelligence functions — using the NASA Task Load Index (NASA-TLX). Welch ANOVAs found no statistically significant differences between the groups on any of the six workload dimensions: mental demand was high across all conditions while physical demand remained low, as expected for a study-based digital task. An exploratory pairwise-weighting analysis revealed two significant differences (mental demand versus perceived performance, and temporal demand versus perceived performance). The author concludes that neither gamification nor AI-supported progression simply increases or reduces workload; rather, cognitive Sustainability depends on how the learning path is designed, paced, explained, and supervised.

Key Findings

1. No statistically significant differences in perceived workload (NASA-TLX) were found across traditional, gamified, and AI-supported learning conditions on any of the six dimensions.
2. Mental demand was high in all conditions; physical demand remained low — consistent with a study-based digital task.
3. Exploratory NASA-TLX pairwise weighting revealed two significant differences involving mental demand versus perceived performance and temporal demand versus perceived performance.
4. Results do not support a simple claim that gamification or AI-supported progression either raises or lowers workload — cognitive sustainability hinges on design, pacing, explanation, and supervision.
5. The study reads its findings alongside NASA-TLX research in professional, industrial, and work-related technology training.

Discussion

The paper contributes a measured, null-result perspective to the often-overenthusiastic literature on gamified and AI-supported learning in higher education. Its value lies in complicating the assumption that adding engagement mechanics or AI-driven progression automatically improves the learning experience — here, workload was not reduced by either condition, and mental demand stayed uniformly high. For the knowledge base, it connects student experience, Motivation, and instructional design literatures, and complements findings on how the quality of technology-mediated environments — not their mere presence — shapes outcomes. It also illustrates the use of standardized measurement instruments (NASA-TLX) to assess the cognitive ergonomics of AI-supported learning.

What this means for practice

• Instructors. Evaluate a gamified or AI-supported unit for perceived workload before scaling it — run a short NASA-TLX check with the class rather than judging it on engagement or outcomes alone.
• Instructors. Expect mental demand to stay high in all conditions and offset it with clear pacing, orientation time, and worked examples instead of adding more game rules.
• Designers. Soften time pressure in constrained pathways: the AI-supported conditioned progression made temporal demand more salient (H = 9.995, p = .007), so explain progression rules up front and avoid hard deadlines inside the loop.
• Instructors. Supervise and explain the AI-supported path — the study's moderate reading is that a structured AI path is tolerable when it is coherent and supervised, not that AI lowers load by itself.

Limitations

• Single-site university sample of 142 students split across three conditions (traditional n = 41; conventional gamification n = 49; AI-supported conditioned progression n = 52), with no claim to generalization beyond this context.
• NASA-TLX is a self-report instrument: it captures perceived workload, not objective cognitive load.
• The analysis isolates the NASA-TLX data from a broader project, so learning outcomes, motivational data, and platform traces are not examined here.
• All six primary comparisons were non-significant, so the interpretation rests on descriptive separations and two exploratory pairwise salience results that were not corrected for multiple comparisons.

Citation

Speranza, M. (2026). Perceived Workload Across Traditional, Gamified and Artificial Intelligence-Supported Learning Conditions: A NASA-TLX Study in Higher Education. EdArXiv preprint.