Theory-guided five-step prompting of AI interaction improves open-ended explanation performance in construction education review, while unprompted AI access alone does not¶

Claim · All claims
Evidence · 1 study · causal r2 · q3
Currency · newest study 2026 · checked against its sources 2026-10-20 (automated, 1 of 1 entry) · not reviewed by a person

Subclaims¶

q3 i? The prompted AI condition had the highest mean open-ended score (12.41, against 9.68 slide-based and 10.45 unprompted), and the omnibus one-way ANOVA across the three conditions was significant (F = 7.32, p = 0.0011); no pairwise post-hoc tests are reported. → Xiaoyu Hou 2026 q3 i? The unprompted AI condition showed no comparable improvement, indicating the benefit is not attributable to AI usage alone. → Xiaoyu Hou 2026

Evidence¶

Xiaoyu Hou 2026¶

Xiaoyu Hou, Bo Xiao, Hexu Liu and Shane Mueller. (2026). The Role of Instructional Guidance in Generative AI-Assisted Learning: Empirical Evidence from Construction Engineering Education. https://arxiv.org/abs/2606.05509 doi:10.48550/arxiv.2606.05509

q3 · i? · causal · r2

Randomized controlled experiment with a final sample of 95 undergraduates (33 slide-based, 29 prompted, 33 unprompted) after screening. Open-ended responses were rubric-scored by two blind raters (ICC = 0.91); the prompted group had a "substantially higher mean score (12.41)", and the one-way ANOVA across the three groups was significant, F = 7.32, p = 0.0011. That is an omnibus test; the article reports no pairwise post-hoc comparisons, so it does not establish that the prompted group differed significantly from each other group individually.

"The Prompted group achieves a substantially higher mean score (12.41) than the NA (9.68) and Unprompted (10.45) groups, with the difference reaching statistical significance (F = 7.32, p = 0.0011)."

Discussion¶

Related Claims¶

• Prompted, unprompted, and slide-based learning conditions yield statistically comparable multiple-choice performance — related
• Prompted AI-supported learning raises overall post-test scores above both slide-based and unprompted AI conditions — related
• Prompting learners to self-explain improves understanding and problem solving on immediate tests by a moderate average amount, with little evidence yet on delayed or classroom gains. — a broader claim this one bears on
• Metacognitive prompts improve learning — a broader claim this one bears on
• Structured prompting raises perceived interaction clarity and dependability but reduces perceived novelty of AI-supported learning — related
• Human tutoring has a medium effect over no tutoring, and in one small study students learned as well when tutors only prompted them as when tutors also explained and gave feedback — related
• Constructive learning beats active and passive learning — related
• Minimal guidance is less effective for novices than explicit instruction — a broader claim this one bears on
• Training on AI interaction strategies does not further improve AI-assisted learning design quality — related
• ARPG+ produces higher final prompt quality than unguided, template, and rule-based conditions among simulated learners — related
• Prompting underspecified responses improves children's performance on both local and global coherence inference questions at all ages — related
• Guided use of a multiple document reading application yields higher multiple document comprehension than access-only, unguided, or traditional pdf viewer use — a narrower finding that bears on this claim
• When explicitly required by direct questions, less skilled comprehenders' inference performance improves greatly while skilled comprehenders perform at ceiling — related
• The tutor-access learning gain was concentrated in conceptual depth of written answers rather than multiple-choice scores — related
• Multi-step AI operations, not AI use alone, are positively associated with creativity of final design work — a narrower finding that bears on this claim