Presentation
Informing AI Prompt Engineering for Flightcrew Training Through Human Factors Research
SessionPoster Session 1
DescriptionAs the aviation industry adopts advanced training technologies, large language models (LLMs) provide new possibilities for delivering adaptive, personalized, and interactive learning experiences for flightcrews. However, the effectiveness of LLMs in this context depends on how well prompts are engineered and crafted to elicit accurate, relevant, and context-aware responses. This study emphasizes the importance of aligning prompt engineering with human factors (HF) principles to ensure compatibility with the cognitive and operational demands of flightcrew training. Using the P.E.R.F.E.C.T. framework, we compare structured prompts with unstructured alternatives across two LLMs—ChatGPT-4o and DeepSeek. Our findings show that HF-informed, context-rich prompts generate higher-quality, domain-specific training content. We further propose integrating Retrieval-Augmented Generation (RAG) to reduce hallucinations and anchor responses in authoritative aviation sources such as FAA manuals and incident databases. Structured prompts were shown to enhance scenario realism and facilitate procedural knowledge transfer. This paper offers practical guidance for flight instructors and instructional designers seeking to incorporate AI into flightcrew training. Future research should investigate multimodal RAG-LLM systems and conduct human-in-the-loop evaluations to ensure AI outputs meet the stringent reliability and safety standards of modern aviation training.
Contributors
Event Type
Poster
TimeTuesday, October 14th5:30pm - 6:30pm CDT
LocationRiverside East