The 5P Reflection Model for Education in the Generative Artificial Intelligence (GenAI) Era

Synthesis: As generative AI makes it hard to authenticate student reflection — a student's reflection may be authored by an AI rather than by the learner — this Design-Based Research paper argues that traditional reflection models (Kolb, Gibbs, DIEP, Schön, Mezirow, Boud) lack strategies for the GenAI era and proposes a GenAI-aware 5P Reflection Model (Purpose, Process, Product, Pitfalls, Plan). The model fuses features from experiential, transformative, reflective-practitioner, emotional, and self-regulated learning theories, embeds explicit handling of GenAI risks (hallucination, plagiarism, over-reliance, Privacy), and embodies a "process over product" philosophy that documents prompts, iterations, and validation rather than just the final output. It positions structured reflection as a way to preserve authenticity, learner agency, and integrity when learners co-create meaning with AI.

Why reflection models must change

Universities worldwide face challenges adopting GenAI because it has affected academic integrity and the scholarship of teaching and research. The paper's core claim is that traditional reflection models are struggling to authenticate student reflection as GenAI becomes incorporated into education. Learners increasingly co-create meaning with AI, so a contemporary reflection model must:

Focus on process rather than product, making the reflective process transparent.
Capture the learner's interactions with AI — the prompts used, and the iterative thinking that unfolds through human-AI collaboration.
Evaluate authenticity by distinguishing learners' own reasoning from AI-generated contributions.
Embed ethical and responsible AI usage — attribution, integrity, disclosure — within the model rather than treating them as external concerns.
Account for the technological mediation of thinking — how algorithms, interfaces, and equitable access shape the reflective experience.

Theoretical foundations of the 5P model

The paper reviews five families of reflection models and analyzes their limitations in the GenAI era:

Experiential models (Kolb, Gibbs, DIEP): valuable for staged, cyclical learning, but GenAI can interrupt learner ownership of reflection and risk abstraction detached from lived experience.
Transformative models (Mezirow): deep re-examination of assumptions, but individualistic and over-emphasizing rational-critical processing.
Reflective-practitioner models (Schön, Brookfield, Rolfe, CARL): useful for reflection-in-action and reflection-on-action, but lack step-by-step structure for documenting AI interaction.
Emotional/affective models (Boud, Moon): attention to feelings, but automation of synthesis can bypass the emotional "stickiness" that genuine reflection requires.
Self-regulated learning models (Zimmerman, Pintrich): goal setting, monitoring, and autonomy, which GenAI can either scaffold or short-circuit.

The five stages

The 5P model is an iterative framework with five stages, each drawing on distinct theoretical features, with emotional factors considered across all stages:

Purpose (self-regulated forethought): learners set clear goals and objectives for both their learning and their intended use of GenAI before any usage — clarifying whether they seek efficiency, Creativity, or clarification, ensuring alignment with institutional and assessment requirements, and guarding against over-reliance driven by anxiety.
Process (reflection-in-action): the active deployment of the tool, documenting the specific prompts and iterations used, monitoring emotional responses (frustration that stops refinement, relief that causes confirmation bias), and validating GenAI's probabilistic output.
Product (reflection-on-action): critically reviewing the final output's quality, originality, and accuracy, cross-validating it against reliable external sources, and distinguishing what the learner contributed versus what GenAI generated.
Pitfalls: an explicit, core stage addressing the inherent risks of GenAI — factual/cognitive (hallucination), ethical/integrity (plagiarism, bias, over-reliance), and privacy/security — and reflecting on where critical human judgment was essential.
Plan: synthesizing insights from all prior stages to set refined goals and a pathway for continuous improvement, driving genuine transformative learning.

Adoption considerations

The model is generally suited to major assessments, capstone projects, or professional-development evaluation rather than routine tasks. Effective implementation requires explicit instruction and Scaffolding from educators — teaching learners to monitor emotional responses, design and refine prompts iteratively, and cross-validate outputs — which itself demands ongoing professional development. The authors note two inherent limitations: assessing the quality of internal thoughts (emotional monitoring, resisting over-reliance) is highly subjective, and documenting the Process stage's prompts and iterations is difficult without specialized tools that capture learners' continuous interaction with GenAI. Overall, the model aims to ensure GenAI provides a structured scaffold rather than a substitute for student thought.

What this means for practice

Instructors. Reserve the model for major assessments, capstone projects, or professional-development evaluations rather than routine tasks, because the five stages require documented check-ins before, during, and after the task.
Instructors. Require learners to write the Purpose-stage goal — whether GenAI is being used for efficiency, creativity, or clarification — before they open the tool, so the "why" of use is captured rather than reconstructed afterward.
Instructional designers. Build a capture mechanism for the Process stage into the assignment (prompt and iteration logs), since that stage is the one the authors say cannot be documented accurately without specialized tooling.
Instructors. Assess the Pitfalls stage explicitly — hallucination, plagiarism, bias, privacy, and over-reliance — so that ethical handling of GenAI is graded rather than assumed.
Researchers. Pilot the model in a GenAI-rich course and compare it against an established reflection framework to test whether the separate Pitfalls stage changes how learners validate AI output.

Limitations

The model is conceptual, developed through design-based research assisted by critical inquiry (a review of Kolb, Gibbs, DIEP, Mezirow, Schön, emotional, and SRL models); no implementation, learner sample, or outcome data are reported.
The authors state that assessing the internal states the Process and Pitfalls stages depend on — emotional monitoring and resisting over-reliance — is highly subjective and relies on the learner's honesty and metacognitive capacity.
The Process stage requires documenting prompts and iterations continuously; the authors state that specialized tools to capture learners' ongoing interaction with GenAI still need to be developed.
Adoption is constrained by design: the authors limit the model to major assessments, capstone projects, or professional-development evaluation, and effective use depends on educators who can already deliver explicit instruction and scaffolding.

Connected Concepts

Connected Articles

Reclaiming Epistemic Agency: A Critical Framework for Human-Generative AI Co-Agency in Education — A critical framework for human–GenAI co-agency in education
How Do Students Regulate Their Learning With a GenAI Chatbot? — How students regulate learning with a genAI chatbot
Scaffold or Shortcut? Postgraduate IT Students' Use of Generative AI and Self-Regulated Learning — Scaffold or shortcut? GenAI's dual role in SRL
GenAI as a Learning Partner: Supporting Self-Regulated Learning Over Time Without Replacing Effort — GenAI as a context-aware learning partner over time
The IDEA Framework for Metacognitively Regulated GenAI Use in Higher Education: Development and Exploratory Pilot Evidence — The IDEA framework for metacognitively regulated GenAI use
Refrain, Then Amplify: A Curriculum Framework for Sequencing Generative AI to Form Professional Judgement — Refrain-then-amplify curriculum framework for sequencing GenAI
A Student-Centered Framework for Responsible Use of Generative AI in Higher Education — Student-facing framework for responsible GenAI use in higher education

Citation

Kadel, R., Shailendra, S., Islam, M. T., Saxena, U. R., Sharma, A., & Kaphle, S. (2026). The 5P Reflection Model for Education in the Generative Artificial Intelligence (GenAI) Era. arXiv:2609.03413.

Metadata

Study design	Theoretical Analysis
Level	Higher Ed
Intended audience	Instructors, Instructional Designers, Researchers
Learning and instruction	Metacognition, Self-Regulated Learning
Technologies and techniques	Generative AI
Research methods and evaluation	Design-Based Research
Foundations of AI in education	Academic Integrity, Learner Agency
Page kind	Framework

Source History Edit Chat