Students might understand how to question artificial intelligence tools but, as deadlines approach, they skip vital steps in artificial intelligence use. Checking AI-generated claims or considering how the tools have shaped their thinking, for example, might fall by the wayside.
I developed the 5P framework for first-year undergraduates in an English-medium instruction (EMI) programme. The framework, paired with brief checkpoints within assignments, offers a way to help students sustain sound AI use practices. Its stages are:
- plan – establish a purpose
- probe – question reasoning and evidence
- pause – examine framing and omissions
- prove – check claims against independent sources
- ponder – reflect on learning and future practice.
A 2026 meta-analysis of 35 studies found benefits from AI-supported self-regulated learning. Importantly, the researchers emphasised that improved performance with AI assistance does not necessarily mean students are developing the ability to regulate their learning independently. The 5P framework encourages them to exercise self-regulation when using AI by turning planning, monitoring and evaluation into practical steps.
In my classroom-based action research, reflections from high-proficiency EMI students suggested that using the 5P framework led to greater awareness of AI’s influence on their thinking. However, students reported using the “prove” and “ponder” stages less consistently under pressure – for instance, when several assignments were due.
- Assessing critical thinking in critical times
- To AI or not to AI? That is the question
- If a chatbot can pass your assignment, you are grading the wrong thing
To counter that, I recommend adding brief checkpoints within assignments to reinforce claim-checking and reflection as workload increases and other classroom scaffolding is withdrawn.
Introduce the framework through a task
Suppose students are preparing an argument on whether university tuition fees should be abolished. One student initially supports abolition and asks AI whether removing fees would make university more accessible.
Before students apply the 5P framework to their own work, educators can model it with this example. Give them a worksheet with a heading for each stage and space for brief notes. The aim is to make students’ decisions about the AI’s output visible: which claims or suggestions they accept, question, check, reject or revise.
Plan: What is my purpose?
Before using AI, students identify what they need to understand, what assistance would be useful and which thinking they should do themselves.
A student’s planning note might read: “I think removing fees could widen access. I’ll use AI to identify other barriers, then evaluate the evidence myself.”
They also decide what would count as a useful, credible response.
Probe: What supports this answer?
Ask students to examine the reasoning and evidence in an example AI response.
AI responds that abolishing fees would increase participation because university would become cheaper. The student questions this reasoning: does a lower price necessarily lead to higher participation? What evidence supports that connection?
Have student pairs identify one part of the answer that needs further explanation, such as a causal link, an unsupported assertion or the treatment of competing arguments. Asking AI to elaborate can help identify what needs investigation but it does not establish accuracy.
Pause: What is missing from this account?
Ask students to examine how the response frames the problem. They might notice an overlooked perspective, an unstated assumption, missing context or an alternative explanation.
The student notices that the answer treats tuition fees as the main barrier. What about students who need childcare or help caring for relatives so they can attend university?
Ask students to record one omission and explain why it matters.
Prove: What can independent evidence establish?
Show students how to check one central AI claim against a reliable independent source.
AI claims that abolishing fees increased university participation. The student finds a study and checks what it shows. Participation may have risen but that alone does not show that removing fees caused the increase. The student revises the claim to match the evidence or leaves it out if they cannot find support.
Ask students to record the source and briefly explain whether its evidence supports, challenges or limits the claim.
Ponder: How did AI influence my thinking?
Students return to their planning note and consider what AI helped them do and which thinking they did themselves.
The student might write: “AI helped me articulate why lower fees could widen access. But following its emphasis on price initially led me to overlook other barriers.”
Ask students what they will retain or change next time. The student decides to examine how an explanation is framed before building an argument around it.
Keep evidence-checking and reflection visible in independent work
When students will be using AI independently, instructors should retain the five questions in the assignment instructions. Give particular attention to “prove” and “ponder”, the stages students reported using less consistently under pressure.
Build two brief checkpoints into the process: a source link with one sentence explaining what the check established, and a short reflection on how AI influenced the student’s thinking and what they would retain or change next time.
In the next suitable class, ask pairs to compare their decisions and unresolved questions, then use a few examples to address recurring difficulties through whole-class feedback.
These brief checkpoints help interrogation of AI outputs and reflection to remain part of students’ practice, even when deadlines loom.
Chris Harwood is associate professor of education at Sophia University, Japan.
If you would like advice and insight from academics and university staff delivered direct to your inbox each week, sign up for the Campus newsletter.
comment