PROCEEDINGS OF THE 12TH GEIA SEMINAR (2025)

This paper summarises a hands-on workshop exploring how aviation English instructors can teach students to make sound decisions about when to rely on artificial intelligence (AI) tools and when not to.The workshop introduces the SAFE framework (Safety, Appropriateness, Feature, Effectiveness) as a practical checklist for evaluating AI use in aviation English instruction.Participants explored text-to-text and speech-to-speech tools, tested prewritten, goalspecific prompts, and reflected on their suitability for independent practice.Findings indicated that detecting hallucinations in tasks more complex than grammatical help or descriptions of aviation concepts demanded a high level of both aviation expertise and English proficiency, generally perceived as at or above ICAO Level 4. This creates a significant challenge: learners who may benefit most from AI support may also be the least equipped to identify its errors or troubleshoot interactions independently.Speech-to-speech tools were considered particularly valuable because of the centrality of speaking and listening in aviation English, although participants identified practical barriers to their independent use.Instructor-authored, pre-vetted prompts emerged as a useful scaffold, while participants saw greater potential for AI at the two ends of the proficiency spectrum: advanced learners and clearly bounded basic practice.The workshop concludes by identifying conditions for safer independent AI use and pedagogical practices that support responsible AI literacy in aviation English education.