Christopher Tsang
P Tsang
Affiliations:
Shalvey College (Christopher Tsang)
Nextus Institute of Science and Technology (P TSANG)
Corresponding author: C TSANG christopher.tsang2@education.nsw.gov.au
This paper examines how contemporary higher education systems unintentionally harm inward learners—students who internalise confusion, distress, and failure rather than expressing it outwardly. Drawing on a detailed case study and institutional document analysis, we show how unclear instructions, rigid policies, mislabelling, and automated administrative delays silently erode self‑esteem, identity, and belonging. The rapid rise of generative AI has intensified these inequalities: students who can safely use AI gain instructional clarity, while risk‑averse inward learners, paralysed by institutional “AI phobia,” are disproportionately harmed.
We introduce a systems‑level framework for understanding inward learners, map the collapse of psychological pillars under ambiguous assessment regimes, and analyse a high‑risk case in which a 30% writing task triggered acute psychological trauma and a missing‑person incident. We then propose ethical reform models and production‑ready AI prompt matrices designed to audit instructional ambiguity, replace cold triage templates with trauma‑informed responses, and restore institutional duty of care. The paper argues that generative AI, when deployed responsibly, can become a tool for educational clarity rather than a new axis of inequality.
Educational ethics has been talking about fairness and wellbeing for years, mostly in glossy policy documents and conference keynotes. What it has not done—at least not with any real precision—is look closely at the students who don’t raise their hands, don’t send the angry email, don’t march into the dean’s office. The ones who go quiet. The ones who take the confusion home and turn it inward. Those are the inward learners. They don’t appear in institutional dashboards. They rarely feature in case studies. Yet they are the ones who absorb every ambiguous instruction, every contradictory policy, every delayed reply as evidence that they themselves are defective. Their collapse is almost always silent, and by the time anyone notices, the damage is already done.
Into this already fragile ecosystem, we have dropped generative AI. Students who feel safe enough to experiment with it—who can risk a misstep without fearing institutional punishment—suddenly gain a powerful ally: clearer instructions, reduced cognitive load, a second pair of eyes on confusing tasks. Inward learners, by contrast, often treat AI as a forbidden object. They have heard the horror stories, the warnings, the rumours. They stay away. This paper sits at that intersection: systemic blind spots and AI misuse. It is not a theoretical exercise. It is written from the vantage point of lived cases, broken semesters, and families who have had to file missing‑person reports because a single assessment document was allowed to be both confusing and compulsory.
If you’ve taught for long enough, you know this student. They sit near the back or the middle, not hiding, but not drawing attention. Their work is often neat. Their emails—if they send any—are apologetic, carefully worded, and usually late. Inward learners do not explode outward when something doesn’t make sense. They implode. They take the assignment home, read it again, and again, and again. They assume the problem is them. If the document contradicts itself, they don’t mark it up and bring it to class; they try to invent a way to obey both halves of the contradiction at once. Psychologically, they are often conscientious, literal, and deeply rule‑bound. They are the students who will not cross a road unless the pedestrian light is green, even at 2 a.m. on an empty street. When confronted with opaque policies or “appropriate yet forbidden” instructions, they experience a kind of cognitive vertigo: every possible choice feels like a potential breach. Without explicit reassurance—“this is confusing, and that’s our fault, not yours”—they quietly absorb systemic defects as proof of their own inadequacy. Over time, the institution’s structural noise becomes their internal voice.
Most universities do not set out to harm these students. The harm happens in the gaps—between policy and practice, between marketing and lived reality. Support systems are built on the assumption that distressed students will signal loudly: they will email, book appointments, lodge appeals, attend workshops. Inward learners rarely do any of those things. They read the support page, see the list of steps, and decide they are not “bad enough” to justify the administrative burden.
Meanwhile, bureaucratic routines quietly raise the threshold for help. Multi‑page forms, long response times, generic auto‑replies, and fragmented communication channels all send the same message: “If you really need us, you will persist.” Inward learners interpret that very differently: “If I am already struggling, I don’t have the stamina to prove I deserve help.” The blind spot is not just emotional; it is architectural. Systems are designed around vocal distress. Silent collapse is treated as an outlier, when in fact it is a recurring pattern.
When you look closely at an inward learner in crisis, you don’t see a single bad week. You see a slow erosion. It starts with small things: an assignment that uses unfamiliar jargon without explanation, a rubric that contradicts the task description, a policy that says “extensions are available” but requires a form that feels like a confession. Each event is survivable on its own. Together, they begin to chip away at self‑esteem and academic confidence. At some point, the narrative shifts. The student stops saying, “This subject is confusing,” and starts saying, “I am not the kind of person who belongs at university.” Identity and belonging are not abstract constructs here; they are the quiet thoughts at 2 a.m. when the student is staring at a PDF that makes no sense and cannot imagine emailing anyone to admit it.
By the time the institution notices—if it notices at all—the pillars have already collapsed. What looks like “sudden disengagement” from the outside is, from the inside, the final stage of a long, unspoken struggle.
One of the cruellest ironies is that inward learners are often described, in staff rooms and progress meetings, as “fine”. They attend. They don’t cause trouble. They don’t send angry emails. When they miss a deadline, the default explanation is time management. When they fail an assessment, the narrative is that they “didn’t read the instructions properly”. Parents, too, can misread the signs. A withdrawn teenager in the middle of a chaotic semester looks, from the outside, like someone who is “just stressed” or “being a typical first‑year”. The early warning signals—hesitation, perfectionism, avoidance—are easy to dismiss.
Diagnostic drift happens quietly. Institutional stories about “students these days” (too dependent on technology, unwilling to read, resistant to feedback) replace careful, individual analysis. Inward learners are folded into these stories, miscategorised as careless or disengaged, when in reality they are paralysed by ambiguity and terrified of doing the wrong thing.
On paper, universities are compassionate, responsive, student‑centred. The brochures are full of smiling faces and reassuring slogans. The wellbeing pages list counselling services, hotlines, and “we’re here for you” messages. The lived reality, especially for inward learners, can be starkly different. Help centres are often physically distant from teaching spaces. Office hours clash with casual work shifts. Email addresses lead to shared inboxes monitored by overworked staff who triage by urgency—and silent distress rarely looks urgent. Automated replies promise responses “within five business days” to students who are already at the edge of their coping capacity. For a student who has internalised the belief that asking for help is itself a kind of failure, these small frictions are not minor inconveniences; they are decisive barriers. The institution’s paper reality says, “We care.” The lived reality, experienced at 11:47 p.m. in front of a confusing assignment PDF, says, “You are on your own.”
Duty of care is one of those phrases universities love to use in policy documents. It sounds noble, protective, almost parental. But in practice, duty of care often collapses into a checklist: did we provide the link, did we publish the policy, did we send the reminder? The deeper ethical obligation — to prevent foreseeable harm — gets lost in the administrative shuffle.
For inward learners, harm is entirely foreseeable. If an institution knows that ambiguous instructions cause distress, and it continues to publish contradictory assessment documents, that is not an accident. It is a structural choice. If it knows that automated triage replies delay help, and it continues to rely on them during high‑risk periods, that is not a neutral workflow. It is an ethical failure. Duty of care is not about whether the university can defend itself in court. It is about whether the institution can look at a collapsed student — or a missing‑person report — and honestly say, “We did everything we reasonably could to prevent this.” Too often, the answer is no.
The arrival of generative AI should have been a turning point for instructional clarity. It could have been the moment academics finally had a tool to check their own documents for ambiguity, contradictions, and cognitive load. Instead, something stranger happened.
Academics embraced AI for research — literature reviews, coding, data cleaning, manuscript polishing — but avoided it almost entirely for teaching. The same professor who uses AI to refine a journal article will still hand students a PDF full of half‑finished sentences, mismatched rubrics, and contradictory instructions.
Meanwhile, institutional AI policies often read like they were written by three committees who never met each other. One paragraph encourages responsible use; the next warns students that AI‑assisted work is “extremely poor quality.” For inward learners, this is not guidance — it is a threat. They avoid AI entirely, even when it could help them understand the task. The blind spot is not technological. It is cultural. Universities have not yet realised that AI is not just a research tool; it is an instructional clarity tool. And inward learners are paying the price for that oversight.
Jenny’s collapse did not begin with the assignment. It began weeks earlier, with a slow accumulation of small confusions she never voiced. But the writing task — the infamous 30% assessment — was the tipping point. The document told her that certain topics were “appropriate” but “must not be used.” It offered no examples of acceptable alternatives. It warned her that AI was permitted, but also that AI‑assisted work was “extremely poor quality” and had caused failures. It demanded a PDF with a rasterised cover sheet to avoid triggering Turnitin. It required a five‑minute video presentation with an absolute failure threshold: miss it, and you fail the entire task. For Jenny, each instruction felt like a trap. She read the document dozens of times, trying to decode what the institution wanted. She did not ask for help. She did not email. She did not attend drop‑ins. She tried to solve the ambiguity alone, because that is what inward learners do. When the fear became unbearable, she left home. Her parents found a cryptic social media post — grammatically perfect, unlike her usual writing — suggesting she was “staying with friend.” They filed a missing‑person report. Police traced her digital footprint. Meanwhile, the university’s automated triage replied to urgent emails with “please wait five business days.”
Jenny returned home eventually, shaken, exhausted, and convinced she had failed not just the task but the entire idea of university. The institution never saw the collapse coming. It never sees collapses like hers coming. That is the problem.
Ethical reform cannot be cosmetic. It cannot be another brochure or another committee. It must begin with the recognition that instructional clarity is a form of psychological safety.
A real framework has three layers: First, institutions must audit their assessment documents the way engineers audit safety systems — looking for contradictions, cognitive load spikes, and ambiguous phrasing that could trigger distress. This is not proofreading; it is harm prevention. Second, communication workflows must be redesigned so that high‑risk messages never receive automated replies. A student in crisis should not be told to wait five business days. Trauma‑informed communication is not optional; it is part of duty of care. Third, AI must be integrated into teaching, not just research. If academics use AI to polish journal articles, they can use it to polish assignment PDFs. If students are expected to navigate complex tasks, they deserve documents that have been checked for clarity. Ethical reform is not about perfection. It is about responsibility.
Institutions often ask for “actionable recommendations,” so here they are — uneven, practical, and grounded in lived cases.
These tools are simple. They are not revolutionary. But they can prevent collapses like Jenny’s. That alone makes them worth adopting.
Inward learners are not rare. They are simply quiet. They do not disrupt classes or flood inboxes. They do not demand extensions or complain about unclear instructions. They absorb the institution’s noise until they break. The tragedy is that their harm is preventable. Clearer documents, humane communication, and responsible AI use could transform their experience. None of these require new buildings, new budgets, or new committees. They require attention — real attention — to the students who do not speak until it is too late. If higher education is serious about fairness, wellbeing, and accountability, it must learn to hear silence. It must learn to see the students who disappear between the lines of its own documents. And it must act before another Jenny walks out the door.