When a student uses AI on an assignment, academic integrity is often the first concern. Was it permitted? Was it disclosed? Does the submission represent the student’s own understanding?
Those are necessary questions. They are also only part of the conversation.
A student may be deliberately avoiding the work. They may also be struggling to interpret the question, organise an argument or write in an academic register they have never been explicitly taught. Understanding that difficulty does not excuse misconduct. It helps us respond to the learning need alongside it.
During an OpenRoads: EdTech Training Caravan session at the University of Southern Philippines Foundation in Cebu City, we brought Social Sciences and Natural Sciences faculty into a conversation about assessment design.
John Carlo, OpenLearning’s EdTech Solutions and Training Specialist, collaborated with the academic leaders and faculty members of the University of Southern Philippines Foundation in Cebu. This collaboration was captured in a photo courtesy of Prof. Mary Joy Morano Sanchez, the EDUTECH Coordinator.
Before discussing frameworks or platforms, we asked: could a student use AI to complete your current assessment without doing the thinking you intended?
The discussion exposed a useful distinction. A task can produce a polished submission without giving the educator enough evidence of how the student reached it. That is a design issue worth examining, even when the assignment has worked well in the past.
A 2025 study published in Higher Education Research & Development followed 24 multilingual undergraduate students who used generative AI for academic writing. What the researchers — Hysaj, Dean, and Freeman — found was not a room full of students looking for shortcuts. What they found were students who were lost.
“I just cannot trust myself to do it on my own,” said one participant.
“Every time I think I understood what the author said, I get it wrong,” said another.
“I don’t understand most of topics and I am desperate to write.”
These accounts suggest that some students needed more explicit support with academic writing. They may have been navigating a hidden curriculum that higher education assumes students already know. That does not excuse academic misconduct, but it gives us another part of the problem to address.
In the Philippine context, this hits differently. For many Filipino students, English is not a first language. Academic writing conventions are not intuitive.
Singer-Freeman et al. (2025) found that students with lower confidence in their writing were significantly more likely to reach for AI. And Fernando et al. (2025), studying Filipino students specifically, found that AI use was shaped by diskarte — the cultural disposition toward resourceful problem-solving. Some students may have been trying to find their way through. That motivation can coexist with academic misconduct.
The pedagogical reframe that came out of our session: when a student uses AI on a reflection paper, the question isn’t only “did they cheat?” The more important question is “did we design an assignment where the thinking couldn’t be outsourced?”
In a Philippine classroom, the practical question is how to support students working across languages, disciplines and different levels of confidence. Do they know what an analytical paragraph needs to do? Can they distinguish summarising a source from evaluating it? Have we shown them what appropriate AI assistance looks like for this particular task?
If the answer is no — and for most of us, at least partially, it is — that’s the work. Alongside clear academic integrity expectations, we need assessment design that makes authentic thinking the path of least resistance.
Clear rules matter. So do worked examples, opportunities to practise and feedback before the final submission. Students need to understand both what is expected and how to get there.
This is a harder conversation than “students are cheating with AI.” It’s also the more honest one. And it’s the one Philippine higher education needs to be having right now.
A useful place to begin is one assignment you already teach. Identify the thinking it is meant to develop, then ask what evidence would make that thinking visible.
For a reflection paper, students could explain an initial view, respond to a classmate’s perspective and describe what changed. For a research task, they could submit a question, an annotated source selection and a short explanation of why they revised their argument. For a science activity, they could interpret their own observations, discuss uncertainty and defend the conclusion they reached.
If AI use is allowed, ask students to explain where they used it, what they checked and what they rejected. Assess the quality of that judgement alongside the work itself. A brief discussion or oral explanation can give students another way to demonstrate their understanding.
None of these approaches makes an assessment immune to misuse. Together, they can give educators a richer basis for feedback and judgement than a final document alone.
The value extends beyond academic integrity. Explaining a decision, evaluating evidence, responding to feedback and improving a piece of work are capabilities students can carry into professional life.
An assessment designed around these practices can help students build examples they can discuss in a portfolio, an interview or a workplace. The connection between learning and opportunity begins with what students are asked to do today.
OpenLearning’s emphasis on social and active learning fits this approach: discussion, peer feedback and reflection can become part of the learning experience. The educational value depends on how faculty design and assess those activities. Technology supports that work; educators give it purpose.
The question is worth taking back to your department: what could we change in one assessment so that students receive better support and we can see more of their thinking?
OpenLearning’s OpenRoads: EdTech Training Caravan brings conversations about AI, assessment and teaching practice to campuses across the Philippines. Start with a 15-minute conversation with John Carlo Santos, Educational Technology Solutions and Training Specialist, to discuss your institution’s assessment challenges and a possible campus session.
Hysaj, A., Dean, B. A., & Freeman, M. (2025). Exploring the purposes and uses of generative artificial intelligence tools in academic writing for multicultural students. Higher Education Research & Development, 44(7), 1686–1700.
Singer-Freeman, K., Verbeke, K., & Barre, B. (2025). Generative AI usage among university students depends on academic level and task. Higher Learning Research Communications, 15(2).
Fernando, E. Q. et al. (2025). Profiles of AI dependency: A latent class analysis of Filipino students’ academic competencies. ICEMT 2025, Osaka.
Bayán-Bayán is OpenLearning’s Philippines-focused editorial series, bringing together reflections on EdTech, AI in education and teaching practice across Philippine higher education institutions. It is a place for the conversations that deserve to travel beyond the room where they began.
Keep the conversation going: subscribe to receive the Bayán-Bayán digest, with reflections, educator stories and practical ideas for Philippine higher education.
