Canonical text (the exact bytes fingerprinted)

TRUVACE RECORD VERSION
record: TRV-2026-2365
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-11-04T12:01:44.005961Z
status: published
lens: g_space
sector: education
headline: VirtuAI-Safe: Personalized Situated Safety Learning through Integration of VR and LLMs
dek: Construction safety outcomes are strongly shaped by human factors, including how individuals perceive hazards, interpret contextual cues, and make safety-related decisions in dynamic work environments. Safety education therefore plays a critical role in preparing future construction professionals to reason about risk and apply safety principles in context. Traditional safety instruction often emphasizes regulatory knowledge and procedural compliance, providing limited support for the development of judgment and…
gain_title: Two days after a one-hour VirtuAI-Safe session, construction management undergraduates performed best on conceptual understanding and knowledge transfer, the judgment skills needed on real sites.
problem_title: (none)
trace_subject: (none)
gain_reading: Two days after a one-hour VirtuAI-Safe session, construction management undergraduates performed best on conceptual understanding and knowledge transfer, the judgment skills needed on real sites.
gain_evidence: Postlearning performance was strongest for conceptual understanding and knowledge transfer. | VirtuAI-Safe effectively supported learning and engagement within the scope of the study
problem_reading: (none)
problem_evidence: (none)
quick_read: Researchers built VirtuAI-Safe, an immersive virtual reality environment that uses large language models to give adaptive, context-aware guidance inside simulated construction scenarios. Undergraduate construction management students used it for a one-hour session on scaffolding safety, and a knowledge test two days later showed the strongest results for conceptual understanding and knowledge transfer.

Construction safety depends heavily on how workers perceive hazards and make decisions in changing conditions, yet traditional instruction centers on regulations and procedural compliance. The study tests whether personalized, situated learning can build the judgment that real sites demand.

Results came from a single-group design with no comparison against conventional teaching, so the findings show what students achieved rather than how much better the tool is than other methods. Factual and procedural scores varied more after the delay, and affective engagement was only moderate.
limitation: The study used a single-group, postlearning design with one one-hour session on one topic, scaffolding safety, so it has no comparison group and its conclusions apply only within the scope of the study.
tag: Evidence-backed gain
key_points: The tool combines virtual reality with large language models to give personalized, context-aware guidance in simulated construction scenarios. | Scaffolding safety was the test domain because it is high-risk and requires interpreting evolving site conditions and multiple safety constraints. | Learners took a knowledge test two days after a one-hour session, covering factual, conceptual, procedural, and transfer knowledge. | Behavioral and cognitive engagement were consistently strong, while affective engagement was more moderate. | Factual and procedural performance varied more after the delay, pointing to a need for reinforcement.
rundown: VirtuAI-Safe was implemented for undergraduate construction management students preparing for professional roles. Instruction is embedded in realistic simulated scenarios rather than delivered as regulatory or procedural content alone.

The evaluation used a single-group, postlearning design. Knowledge was tested two days after a one-hour session across four types: factual, conceptual, procedural, and transfer. Engagement was measured right after the session on behavioral, cognitive, and affective dimensions.

Conceptual understanding and transfer scored highest, while factual and procedural knowledge were more variable after the delay. The authors say this suggests reinforcement is needed to sustain detailed recall and stepwise execution, and that the findings can inform future adaptive instructional support in immersive learning systems.
sources:
- peer_reviewed | Journal of Computing in Civil Engineering | https://doi.org/10.1061/jccee5.cpeng-8010 | 2027-01-01
prev: 0000000000000000000000000000000000000000000000000000000000000000

sha256

20b21b6ebd98fa20b59ae69209342a5ed8344ce7d085cb6a0d947856f69f9841

previous

0000000000000000000000000000000000000000000000000000000000000000