In the ever-evolving intersection of education and artificial intelligence, prompt engineering (PE) has emerged as a critical skill and scientific methodology that powers personalised, adaptive learning experiences. Drawing on recent academic frameworks and grounded in learning analytics (LA), prompt engineering is no longer just an art of writing clever commands for chatbots it is the foundation of intelligent, learner-responsive education systems.

This article synthesises insights from current research and practice to demonstrate how prompt engineering is reshaping digital learning, and how it can be operationalised within educational technology design.

What Is Prompt Engineering in Education?

Prompt engineering refers to the process of designing input instructions that guide generative AI (G-AI) systems – like ChatGPT – to produce specific, meaningful, and educationally relevant outputs. In the context of learning, PE acts as the translator between raw learner data and intelligent, tailored educational interventions.

Rather than static content delivery, prompt engineering enables dynamic, data-informed conversations that adjust in real-time to the learner’s needs.

The Feedback Loop: Learning Analytics meets AI

At the heart of effective prompt engineering in education lies a powerful feedback loop:

This loop is the crux of a new model of adaptive learning. Learning analytics – tracking metrics like engagement, performance and misconceptions – feed into the prompt design process. For example, if a student shows confusion in a topic, the AI can be prompted to generate simpler analogies or more visual explanations. Conversely, high performance might trigger advanced challenges or enrichment tasks.

These prompts aren’t generic; they are designed with intention, using structured frameworks to ensure precision, relevance and learner engagement.

Frameworks that shape prompts

Academic research has introduced several structured approaches to prompt design. Among the most influential are:

• CRISPE: Assigns a clear role to the AI, provides context (Insight), sets the task (Statement), modulates tone (Personality) and invites iteration (Experiment).
• PLAF (Prescriptive Learning Analytics Framework): Guides the use of learner data to structure prompts dynamically and ethically.
• CLEAR & AIPROMT: Emphasise clarity, specificity, logical sequencing and iterative refinement.

These frameworks transform PE from guesswork into a replicable, evidence-based practice. They also support prompt templates and few-shot learning techniques, enabling consistent performance even in complex educational contexts.

Practical use cases: Adaptive Learning in Action

Here’s what this looks like in practice:

This responsive model allows educators to implement AI systems that listen, adapt and respond – mirroring effective human tutoring but at digital scale.

Beyond automation: A new literacy for educators

Research highlights the importance of training educators and students in prompt engineering itself. PE isn’t just a backend process – it’s a skill. Teaching prompt design empowers educators to shape AI behaviour and supports students in becoming co-creators of their own learning paths.

Incorporating prompt engineering into AI literacy curricula is a step toward sustainable, ethical and inclusive educational AI use.

Challenges and the road ahead

While the potential is vast, challenges remain. The subjective nature of evaluating AI output, evolving model capabilities and the need for rigorous validation are key concerns. Future research aims to:

• Develop automated prompt quality metrics
• Explore prompt engineering at scale across diverse learner groups
• Refine ethical use of learner data in real-time prompting
• Embed PE into educator professional development globally

From theory to transformation

Prompt engineering is not a peripheral tech skill – it is surely central to the next generation of educational design. By linking structured frameworks with live learner analytics, educators can create adaptive, engaging and deeply personalised learning experiences.

My ongoing PhD research seeks to advance this vision, establishing prompt engineering as a rigorous, data-driven methodology capable of transforming how we teach and learn with AI.

References & further reading

• Lee & Palmer (2025): Systematic review on prompt design frameworks in HE
• Chen et al. (2022): AdaPrompt – adaptive model training for NLP
• Wang et al. (2025): A Sequential Optimal Learning Approach to Automated Prompt Engineering in Large Language Models
• Knoth et al. (2024): AI literacy and its implications for prompt engineering strategies
• Henrickson & Meroño-Peñuela (2025): Prompting meaning: a hermeneutic approach to optimising prompt engineering with ChatGPT

Damien Caldwell
inico DIGITAL