Teaching higher vocational students to evaluate AI career advice: TRACE scaffolding and appropriate reliance in a cluster-randomised trial

Career guidance requires students to judge whether AI recommendations are credible and suitable to their circumstances. We tested TRACE, a decision routine practised with progressively reduced prompts, in a teacher-blocked 2 × 2 cluster-randomised trial involving 480 second-year students in 24 classes at three Chinese higher vocational institutions. The focal comparison was AI + TRACE versus AI only. On a 10-item assessment, appropriate reliance was 81.3% versus 72.1%. The class-adjusted difference was 9.06 percentage points (pointwise randomisation interval [2.54, 24.12]; exact p = .031; Holm-adjusted p = .063), with all six teacher-block differences positive. The mixed-model odds ratio was 1.68 (CR2 95% CI [1.30, 2.16]). Early unaided transfer was 1.14 points (Holm-adjusted exact p = .094), while delayed performance showed little separation. TRACE improved evaluation of sampled AI career advice, but delayed unaided transfer was not established.

Authors

Institutions

Publication Details

Journal
British Journal of Guidance and Counselling
Published
2026-11-13
DOI
https://doi.org/10.1080/03069885.2026.2744378
Primary Topic
Artificial Intelligence in Education
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Teaching higher vocational students to evaluate AI career advice: TRACE scaffolding and appropriate reliance in a cluster-randomised trial

Yushuang Wu, Haohong Nie, SENHAO XIONG, Kuihua Li
British Journal of Guidance and Counselling
Artificial Intelligence in Education
article

Teaching higher vocational students to evaluate AI career advice: TRACE scaffolding and appropriate reliance in a cluster-randomised trial

Yushuang Wu, Haohong Nie, SENHAO XIONG, Kuihua Li
article en

Abstract

Career guidance requires students to judge whether AI recommendations are credible and suitable to their circumstances. We tested TRACE, a decision routine practised with progressively reduced prompts, in a teacher-blocked 2 × 2 cluster-randomised trial involving 480 second-year students in 24 classes at three Chinese higher vocational institutions. The focal comparison was AI + TRACE versus AI only. On a 10-item assessment, appropriate reliance was 81.3% versus 72.1%. The class-adjusted difference was 9.06 percentage points (pointwise randomisation interval [2.54, 24.12]; exact p = .031; Holm-adjusted p = .063), with all six teacher-block differences positive. The mixed-model odds ratio was 1.68 (CR2 95% CI [1.30, 2.16]). Early unaided transfer was 1.14 points (Holm-adjusted exact p = .094), while delayed performance showed little separation. TRACE improved evaluation of sampled AI career advice, but delayed unaided transfer was not established.

British Journal of Guidance and Counselling
Khon Kaen University (TH), Chongqing Three Gorges University (CN)
Artificial Intelligence in Education
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.

Laika access coming soon