Prompt engineering as a key factor in creating pedagogically relevant tasks using artificial intelligence

this article explores the role of prompt engineering in creating pedagogically relevant tasks using artificial intelligence. The authors highlight the gap between the potential of generative neural networks and the lack of competencies among prospective teachers to interact with them. Through interdisciplinary research, it was experimentally proven that simple (Zero-shot) prompts show low effectiveness (3.42 points). In contrast, the highest effectiveness (4.72 points) was demonstrated by a combined approach that integrates role-framing with Few-shot and Chain-of-Thought techniques. Based on these results, a structured six-component algorithm for prompt design was developed, which includes defining the AI's role, specifying the pedagogical task, and requesting a step-by-step rationale. In conclusion, the article proposes practical recommendations for implementing prompt engineering into teacher training systems, including the development of educational modules, workshops, and updating the professional standard for educators.

generative artificial intelligence, prompt engineering, pedagogical design, educational tasks, Chain-of-Thought, Few-shot, educational technologies