How to prompt AI tutors: New research shows the technique that actually works
Researchers have identified how to engineer prompts that make generative AI effective for teaching—a finding with immediate implications for EdTech companies and schools deploying AI-powered learning tools. The work reveals that aligning prompts with sound learning theory, rather than trial-and-error approaches, dramatically improves student knowledge retention and engagement.
Originaltitel: Knowledge construction with GenAI: The role of theory-informed prompt engineering in achieving pedagogical alignment
<p>Although generative artificial intelligence (GenAI) tools are increasingly integrated into education, limited attention has been paid to how prompt design mediates pedagogical alignment and instructional intent. This study addresses this gap by exploring the interplay between theory-informed prompt engineering and the coherence of generated content with curricular goals. We employ a theory-driven prompt engineering approach based on the Social Knowledge Construction framework and evaluate the pedagogical and technological alignment of ChatGPT-generated content using the Technological Pedagogical Content Knowledge (TPACK) framework. We analyse ChatGPT’s responses across five instructional aims: organising teaching activities, integrating technology, assessment, student engagement and meeting diverse student needs. For this, a structural equation model was developed through which we also analysed the relationship among TPACK dimensions in the GenAI-generated outputs. Results demonstrate that as theoretically informed prompts increase the quality of ChatGPT-generated outputs in terms of pedagogical alignment with learning goals that are reflected in TPACK dimensions.</p>