Scaffolding Futures: Teaching Critical Engagement with LLMs for Foresight in Leisure Education

Abstract: This paper reports on an educational approach designed by the authors that teaches students to critically engage with Large Language Models (LLMs) for future research in leisure domains. It explores how structured scaffolding techniques can transform LLM hallucinations from a limitation into a creative asset for foresight practice, while positioning students as responsible architects of AI-assisted futures exploration. The methodology combines scaffolding techniques for LLM interaction with Bevolo and Draeger's approach to creative hallucination management in foresight practice. The paper presents a five-step prompting framework that guides students through systematic exploration of leisure futures, from setting contextual foundations to synthesizing coherent scenarios. This approach was implemented within a major generalist program about Leisure Futures in a workshop format where students generated individual "signals" to be clustered and mapped onto quadrants to identify emerging trends. Preliminary results from trials with students demonstrate that the structured scaffolding approach enables them to direct LLM outputs more effectively. Rather than using simple, open-ended questions, students learned to analyze problems, deconstruct them, and engage with LLMs in a structured manner that gave them greater control over outcomes. Practical Implications: The developed framework offers educators a practical and replicable model for teaching responsible AI usage while fostering critical thinking. The approach is particularly valuable for foresight, futures research, and design futures education, where creative exploration of future scenarios based on “signals” with future potential relevance is mission critical. Keywords: AI-augmented foresight; constructive LLM hallucination; forward-looking design methods; prompt-scaffolding methodology; responsible GenAI facilitation; systematic futures inquiry.