Abstract

The rapid integration of artificial intelligence (AI) into higher education has generated widespread institutional anxiety, largely addressed through technological surveillance and punitive anti-cheating frameworks. This position paper challenges the dominant narrative by arguing that the erosion of educational quality stems not from AI itself, but from the structural disarray of the contemporary university. Escalating workloads, compressed deadlines, and the fragmented tempo of academic life—where lectures, laboratories, assessments, and administrative demands frequently converge—undermine students’ capacity for sustained intellectual engagement. Within such conditions, delegating cognitive and procedural tasks to AI becomes less an expression of laziness than a rational strategy of academic survival, transforming learning into a model of quasi-productivity characterized by transactional “submit-and-forget” practices.

Moreover, this form of digital alienation extends beyond students to faculty members, who increasingly operate within screen-mediated Learning Management Systems (LMS) that privilege metric surveillance over meaningful human interaction.

While hypertext historically revolutionized knowledge navigation, modern AI-mediated interfaces risk conflating the mere illusion of information access with true comprehension. Drawing on empirical observations of student attitudes toward intellectual autonomy and educational responsibility, this paper advocates a shift away from algorithmic regulation toward AI-resilient pedagogical design. Ultimately, we contend that sustainable educational innovation does not require abandoning pedagogical heritage; rather, universities must recover enduring traditions of structured academic pacing, intellectual continuity, and face-to-face dialogic mentorship in order to reclaim higher education as a space for authentic human thought.

1 Introduction

The rapid proliferation of generative artificial intelligence (AI) in higher education has triggered widespread anxiety at universities, primarily focused on academic dishonesty and the erosion of student integrity. Universities worldwide have responded with reactive regulation strategies, rushing to implement AI detectors and strict prohibition policies. This policing mindset spans the entire academic hierarchy: it is equally visible in the swift ban-and-detect mandates of elite Western institutions like Oxford and Harvard, the rigid policy adjustments at major local public institutions like the University of Warsaw, and the digital monitoring frameworks adopted by specialized private universities like the University of Information Technology and Management (UITM) in Rzeszów.

However, this mainstream discourse treats AI adoption as an isolated moral failure of the student, completely ignoring the structural environment of modern higher education. Contemporary universities are locked in a systemic race to demonstrate measurable achievements, driven by international rankings [3, 4, 5, 10], institutional prestige [14, 16] and the aggressive pursuit of external funding streams[15]. When success is defined strictly by an institution's ability to compete and maintain influence on multiple fronts, education ceases to be a space for deep reflection—it becomes a high-speed production system where what matters is only what can be measured, counted, and compared.

This macro-level pressure trickles down directly into the classroom, directly impacting the education system's interaction with students. Contemporary university curricula are increasingly characterized by growing assignment volumes and shrinking deadlines [2, 13, 18] which serve as a false indicator of "high standards" and create a mere illusion of academic rigor. Thus, universities themselves create unbearable conditions. In this high-pressure learning environment, the unmanageable volume of bureaucratic ’busywork’ leaves students with a critical deficit of time. Consequently, the traditional pathway to academic success—characterized by deep cognitive engagement, trial and error, and slow reflection—becomes a luxury that students can no longer afford.

When faced with overlapping deadlines and unrealistic performance metrics, even highly motivated students are forced to adopt tactical survival mechanisms. In this context, outsourcing assignments to generative AI becomes a rational coping strategy rather than a sign of laziness. Tools like ChatGPT can effortlessly generate essays, data, and graphics with unprecedented speed, allowing students to bypass traditional writing, grammar, and prior knowledge barriers. This forced reliance shifts the educational paradigm from actual knowledge assimilation to quasi-productivity—the rapid, effortless generation of superficial textual outputs. Because the university system rewards results and deadlines more than genuine learning outcomes, AI does not merely "spoil" learning; it rewards a mutually beneficial pseudo-productivity where formal passing rates are met, but the genuine quality of education is severely compromised.

For the scholarly community professionally engaged with hypertext [7, 17] this widespread automation raises broader questions concerning authorship [12, 21] the delegation of cognitive functions [20] the evolving relationship between humans and textual systems [11, 22] and the role of the teacher in assessing automated outputs [8, 9]. Yet, the root of the crisis remains unaddressed.

While hypertext historically transformed how users navigate content, organize memory, and access information, generative artificial intelligence is fundamentally reshaping this dynamic. Today, AI mediates knowledge acquisition in ways that risk conflating the mere illusion of access with true comprehension.

This provocation paper argues that the current decline in educational quality is not driven by the existence of AI itself, but by overburdened university curricula that systematically encourage using AI-driven shortcuts [1, 19]. To prevent total cognitive de-skilling and restore deep learning [6] universities must shift their focus from monitoring and regulating technology to optimizing academic workloads. The aim of this provocation paper is to analyze the mechanisms of forced AI-outsourcing, evaluate the illusion of competence it creates, and propose alternative, AI-resilient assessment frameworks to prevent pseudo-productivity.

2 The Disruption of Academic Rhythm

The core issue of the modern educational environment extends beyond heavy workloads; it lies in the systemic collapse of structural time management. Historically, higher education maintained a clear, linear division of the semester into distinct phases: instructional weeks (lectures and practical/laboratory classes), followed by a dedicated credit/attestation week, and culminating in a rigorous examination period that allocated several days of uninterrupted self-guided preparation per subject. Today, this vital academic rhythm has been replaced by a chaotic, fragmented schedule. It is increasingly common for students to attend a theoretical lecture in one discipline, complete a complex laboratory assignment in another, and sit for a high-stakes exam in a third—all within the span of a single day.

This institutional fragmentation completely ignores the cognitive nature of intellectual labour. Deep comprehension and critical analysis are inherently non-linear processes; one cannot schedule a fixed number of insights per minute or predict a precise hourly quota for conceptual breakthroughs. True intellectual work demands open-ended, slow reflection—a mental spaciousness that the current ’all-at-once’ schedule directly discriminates against. When students realize that an assignment requires an unpredictable and substantial investment of time, yet they are forced to shift focus immediately to an upcoming exam, a intense psychological anxiety emerges. The primary objective shifts from authentic learning to immediate relief from a burdensome academic obligation. In this survival mode, outsourcing tasks to generative AI becomes the only viable escape option.

Consequently, this environment institutionalizes a counterproductive ’submit-and-forget’ (transactional) learning model. Students use AI to instantly bridge the gap between chaotic schedules and high university demands. The immediate metric is achieved—the assignment is submitted; the grade is recorded—but the knowledge is instantly evaporated. The most devastating long-term consequence of this pattern is the failure to develop the habit of independent, systematic, and self-directed lifelong learning, which is critical for professional success in the modern working world. Crucially, this systemic failure cannot be blamed on student indolence. Students are not rejecting the value of education; rather, they are rationally adapting to a poorly organized university structure that makes authentic, deep learning impossible because of intrinsic features of this structure.

3 Forced AI-Outsourcing as a Rational Adaptation Strategy

Under the weight of curriculum fragmentation and collapsed academic rhythms, the deployment of generative AI shifts from an act of academic misconduct to a well-thought-out strategy of rational adaptation. The concept of bounded rationality can be applied to demonstrate the internal logic of students: they strive to maximize the critically limited resource of cognitive energy and time. When a university demands the simultaneous execution of fundamentally different academic tasks within an impossible time-frame, the traditional ideal of the deep learning breaks down. The student is forced to perform a pragmatic cost-benefit analysis. In this equation, outsourcing execution to AI seems the acceptable mechanism that prevents total burnout while ensuring formal compliance with university metrics.

And most importantly, this systemic pressure provokes a profound ’paradox of intention,’ disproportionately affecting intrinsically motivated students. These individuals enter higher education with a genuine desire for deep subject proficiency and personal growth. However, when confronted with overlapping deadlines and the chaotic combination of lectures, labs, and exams, their intrinsic motivation turns against them. They face an insolvable dilemma: either sacrifice sleep and mental health to complete all tasks in good faith, or pragmatically delegate the labor to AI. By choosing the latter, even excellent students are forced to separate their education—using AI to absorb the unwanted expenditure of time and effort on secondary assignments so they can preserve their remaining cognitive capacity for truly useful sections of the curriculum. Thus, the university systematically forces students into a corner, transforming active and dedicated learners into hostages of circumstance who love to learn but are forced to manoeuvre among strict rules.

In this distorted learning environment, generative AI functions not as a tool for intellectual relaxation, but as a survival mechanism and a cognitive equalizer. It bridges the structural gap between human cognitive limits and unrealistic university expectations. By accelerating text production and automating data processing, AI allows students to simulate the hyper-productivity that universities demand with increasing pressure. However, this optimization comes at a high price. Outsourcing the solution of a problem inevitably removes the student from the most valuable stage of the learning cycle: overcoming uncertainty and consistently synthesizing ideas. As a result, the rational choice to survive the semester unintentionally accelerates the erosion of the very nature of education the student strives to acquire.

4 The Mutual Degradation Loop: Fragmented Assessment and the Illusion of Competence

The architectural fragmentation of the modern university does not merely entrap the student; it operates as a two-sided trap that equally compromises the faculty. Professors find themselves trapped in an exhausting cycle, constantly switching between multiple unrelated disciplines, diverse teaching formats, institutional administrative burdens, and research projects. This structural overload is further exacerbated by the omnipresence of Learning Management Systems (LMS). Rather than facilitating meaningful interaction, the LMS forces the faculty into the role of data monitors and managers, binding them to screens to track and assess student activity and metrics. In this environment, the thorough, deep evaluation of student work becomes virtually unfeasible. Overburdened faculty are systematically forced toward superficial and accelerated assessment (scanning for keywords, checking final results without considering the method, etc.) and introduction of computer-based tests (checking boxes).

This algorithmic, LMS-driven environment catalyses a systemic crisis of educational identity, leaving students with a chronic sense of institutional alienation. Genuine higher education is traditionally rooted in dialogic mentorship—the living intellectual exchange where a student is recognized as a developing individual, not a quantifiable data profile. When this human connection is replaced by digital interfaces and fragmented schedules, the motivation for real academic endeavour dissipates. If the university views the student merely as an element of digital data, the student logically responds by viewing the institution through a purely transactional lens. In this formalized system, the desire to engage in slow, creative, and personalized intellectual work is replaced by a pragmatic urge to satisfy institutional metrics with maximum efficiency, with generative AI serving as the ultimate tool for automated compliance.

However, the assumption that generative AI operates as a universal equalizer for academically disengaged students is challenged by empirical reality. Our observations of a digital examination session on the Moodle platform reveal a distinct cognitive stratification. While intrinsically motivated students experience acute ethical conflict—frequently electing formal academic failure and subsequent remediation over technological outsourcing to accurately map their actual competence—disengaged students often fail despite unrestricted access to AI tools. This paradox exposes a critical cognitive barrier: the effective utilization of generative AI requires a pre-existing conceptual framework. Individuals who lack foundational knowledge are fundamentally incapable of formulating precise prompts or executing the evaluation necessary to verify the AI output. As one student noted, ’they simply do not know how to ask AI.’ Consequently, institutional anxieties regarding universal, AI-driven academic success are overstated; the technology does not salvage deficient academic effort, but instead may facilitate a cynical form of credentialism—allowing a segment of students to secure qualifications without competence—while simultaneously compromising the creative capacity and skill acquisition of highly motivated learners.

5 Discussion: From Algorithmic Regulation to AI-Resilient Pedagogical Architecture

The dominant institutional response to the rise of generative AI—characterized by technological surveillance, AI detectors, and punitive policies—is fundamentally flawed. This reactive approach treats symptoms rather than the root cause, further exacerbating student alienation and transforming the educational space into a battlefield of mutual distrust. If universities continue to design assessments that an AI tool can execute with greater speed, efficiency, and less effort than a human, they implicitly concede that their curricula are misaligned with contemporary realities. Therefore, the strategic objective of higher education must shift from regulating technology to redesigning tasks so that they explicitly require and validate unique human agency.

To foster intellectual autonomy, universities must move away from volume-based, transactional grading metrics toward AI-resilient assessment frameworks. The growing scepticism expressed by some students—crystallized in the focus group question, ’Why assume that AI can do it better than me?’—highlights a critical pedagogical opportunity. This mindset represents a baseline of intellectual confidence that universities must actively nurture. Assessments should be structured not to extract a polished final product, but to capture the subjective, non-linear human process of creation. This can be achieved through oral defence formats, live collaborative problem-solving, and reflective learning journals where students map their individual cognitive breakthroughs, errors, and personal evolution.

Crucially, this shift requires active collaboration with the creators of our digital environments. The Hypertext community must help design hypermedia tasks and interfaces that teach critical co-authorship with machines—ensuring AI enhances learning rather than simply accelerating outcomes at the expense of independent thought. As students increasingly transition into supervisors, editors, and verifiers of machine-generated content, we face a critical ethical dilemma: can they truly be considered the authors of this work?

Furthermore, reforming learning architectures requires the restoration of a coherent academic rhythm. Universities must re-establish clear boundaries between instructional phases and dedicated periods of reflective evaluation, thereby reducing the fragmented task-switching that encourages students to outsource cognitive labour to AI systems. At the same time, institutions should counterbalance the rigid, screen-mediated isolation reinforced by contemporary Learning Management Systems (LMS). By minimizing bureaucratic busywork for both faculty and students, universities can reclaim the physical and temporal conditions necessary for sustained dialogic mentorship. Re-centering education around human relationships ensures that students are recognized as developing individuals rather than data points, rendering the accelerationist logic of generative AI structurally irrelevant to meaningful academic growth.

6 Conclusion

In conclusion, the current decline in educational quality and the growing reliance on AI-driven shortcuts are not the direct consequences of technological disruption, but rather symptoms of a deeper structural crisis within contemporary higher education. Generative AI has merely exposed and amplified existing institutional weaknesses, including unsustainable workloads, fragmented academic schedules, and the metrics-driven isolation reinforced by Learning Management Systems. When universities reward rapid content production and procedural compliance instead of sustained intellectual engagement, they unintentionally reshape students from independent thinkers into tactical survivors who rely on AI tools simply to navigate an increasingly unmanageable academic environment.

However, empirical evidence also provides substantial grounds for optimism. As demonstrated by students who deliberately choose self-assessment over automated shortcuts, or who critically challenge the presumed superiority of AI-generated outputs, the human aspiration for authentic intellectual ownership remains remarkably resilient. Such examples suggest that higher education does not require more aggressive forms of technological surveillance or control. Rather, it requires an institutional environment that acknowledges limits of human attention and cognition, cultivates intellectual confidence, and restores meaningful space for face-to-face dialogic mentorship.

Ultimately, this paper serves as a critical appeal to reconsider the trajectory of contemporary educational reform. Innovation in higher education is undeniably necessary, but only insofar as it advances intellectual development rather than becoming a bureaucratic objective designed primarily to satisfy ministerial reporting requirements and digital performance metrics. Genuine progress does not demand the wholesale rejection of pedagogical tradition. On the contrary, enduring academic practices—such as structured educational rhythms, dedicated examination periods, and meaningful personal engagement between faculty and students—have become more valuable than ever in the age of artificial intelligence. If the university is to endure and flourish in this new era, it must reconnect with its foundational principles and reorient its institutional architecture around the slow, profound, and imperfect yet creative character of human cognition.

Source

ACM full-text HTML

ACM Digital Library

Imported from ACM’s structured HTML source. ACM Reference Format: Małgorzata Rataj and Iryna Berezovska. 2026. The Efficiency Trap: Fragmented Curricula and AI-Driven Surface Learning in Higher Education. In 37th ACM Conference on Hypertext (HT '26), September 14--18, 2026, London, United Kingdom. ACM, New York, NY, USA 4 Pages. https://doi.org/10.1145/3800935.3830885

References

[1] Vladymir Azarov and Vladymir Gorokhov. 2025. Cognitive Offloading or Atrophy? Risks of Skill Degradation and New Forms of Dependence in the Era of Pervasive Artificial Intelligence. In International Conference on Quality Management, Transport and Information Security, Information Technologies (QM&TIS&IT). IEEE, Kaliningrad, Russian Federation, 166–168. https://doi.org/10.1109/QMTISIT67022.2025.11468467

[2] Arnold B. Bakker and Karina Mostert. 2024. Study Demands–Resources Theory: Understanding Student Well-Being in Higher Education. Educational Psychology Review 36 (2024), 92. https://doi.org/10.1007/s10648-024-09940-8

[3] Christiane Barats. 2020. Dissemination of International Rankings: Characteristics of the Media Coverage of the Shanghai Ranking in the French Press. Palgrave Communications 6, 1 (2020), 1. https://doi.org/10.1057/s41599-020-0440-5

[4] Gary Barron. 2022. How University Rankings Are Made Through Globally Coordinated Action: A Transnational Institutional Ethnography in the Sociology of Quantification. Higher Education 86 (2022), 809–826. https://doi.org/10.1007/s10734-022-00903-y

[5] Jelena Brankovic, Julian Hamann, and Leopold Ringel. 2023. The Institutionalization of Rankings in Higher Education: Continuities, Interdependencies, Engagement. Higher Education 86 (2023), 719–731. https://doi.org/10.1007/s10734-023-01018-8

[6] Eriona Çela, Mathias Mbu Fonkam, and Rajasekhara Mouly Potluri. 2025. Risks of AI-Assisted Learning on Student Critical Thinking: A Case Study of Albania. International Journal of Risk and Contingency Management 12, 1 (2025), 1–19. https://doi.org/10.4018/IJRCM.350185

[7] Jason Kit Tham Chew and Bob Grace. 2020. Reading Born-Digital Scholarship: A Study of Webtext User Experience. Computers and Composition 58 (2020), 102601. https://doi.org/10.1016/j.compcom.2020.102601

[8] Debby Cotton, Peter Cotton, and Reuben Shipway. 2024. Chatting and Cheating: Ensuring Academic Integrity in the Era of ChatGPT. Innovations in Education and Teaching International 61, 2 (2024), 228–239. https://doi.org/10.1080/14703297.2023.2190148

[9] Nassim Dehouche. 2021. Plagiarism in the Age of Massive Generative Pre-Trained Transformers (GPT-3). Ethics in Science and Environmental Politics 21 (2021), 17–23. https://doi.org/10.3354/esep00195

[10] David D. Dill and Maarja Soo. 2005. Academic Quality, League Tables, and Public Policy: A Cross-National Analysis of University Ranking Systems. Higher Education 49, 4 (2005), 495–533. https://doi.org/10.1007/s10734-004-1746-8

[11] Rebecca Jean Emigh. 2024. Whither Digitality? The Relationship Between Orality, Literacy, and Digitality, Past and Present: From Spoken Traditions to Digital Media. Annual Review of Sociology 50 (2024), 715–736. https://doi.org/10.1146/annurev-soc-033022-035644

[12] Mohammad Hosseini, Lisa M. Rasmussen, and David B. Resnik. 2024. Using AI to Write Scholarly Publications. Accountability in Research 31, 7 (2024), 715–723. https://doi.org/10.1080/08989621.2023.2168535

[13] Balázs Jagodics and Éva Szabó. 2023. Student Burnout in Higher Education: A Demand–Resource Model Approach. Trends in Psychology 31 (2023), 757–776. https://doi.org/10.1007/s43076-021-00137-4

[14] Camille Kandiko Howson, Kelly Coate, and Tania de St Croix. 2018. Mid-Career Academic Women and the Prestige Economy. Higher Education Research & Development 37, 3 (2018), 533–548. https://doi.org/10.1080/07294360.2017.1411337

[15] Vuokko Kohtamäki. 2023. Strategic Dependence on External Funding in Finnish Higher Education. Cogent Education 10, 2 (2023), 2282816. https://doi.org/10.1080/2331186X.2023.2282816

[16] Marek Kwiek. 2021. The Prestige Economy of Higher Education Journals: A Quantitative Approach. Higher Education 81, 3 (2021), 493–519. https://doi.org/10.1007/s10734-020-00553-y

[17] Timothy Lockridge, Enrice Rodrigo, and Cynthia Johnson. 2017. The Kairos Preservation Project. Computers and Composition 46 (2017), 72–86. https://doi.org/10.1016/j.compcom.2017.09.002

[18] Daniel J. Madigan and Thomas Curran. 2021. Does Burnout Affect Academic Achievement? A Meta-Analysis of Over 100,000 Students. Educational Psychology Review 33, 2 (2021), 387–405. https://doi.org/10.1007/s10648-020-09533-1

[19] Fakhroddin Noorbehbahani and Kiemute Oyibo. 2026. AI-Overdependence and Human Cognitive Decline: Hazards, Evidence, and Mitigation Strategies. Computers in Human Behavior Reports 22 (2026), 101102. https://doi.org/10.1016/j.chbr.2026.101102

[20] Simon Rowberry. 2025. The Early ACM Hypertext Conference's Role in Developing Pre-Web Reading On-Screen. New Review of Hypermedia and Multimedia 31, 3 (2025), 175–190. https://doi.org/10.1080/13614568.2025.2477123

[21] Chris Stokel-Walker. 2023. ChatGPT Listed as Author on Research Papers: Many Scientists Disapprove. Nature 613, 7945 (2023), 620–621. https://doi.org/10.1038/d41586-023-00107-z

[22] Kok-Sing Tang. 2025. AI-Textuality: Expanding Intertextuality to Theorize Human-AI Interaction with Generative Artificial Intelligence. Applied Linguistics XX (2025), 1–19. https://doi.org/10.1093/applin/amaf016