College instructors across the country are overhauling their curricula as generative artificial intelligence tools render traditional homework assignments obsolete. Educators report that students now use 'agentic' LLM browsers to complete entire courses of formative assessments in seconds, effectively bypassing the intended learning process.
This shift forces a difficult choice for faculty: continue offering valuable learning opportunities and accept widespread cheating, or remove those assignments entirely. The latter option, while protecting academic integrity, strips students of vital practice.
The erosion of asynchronous learning
Many departments are reverting to oral exams or supervised, handwritten tests to circumvent AI-assisted cheating. However, these methods remain difficult to implement in asynchronous online programs. These courses serve as a lifeline for students with physical disabilities, those living in remote areas, and working professionals balancing caregiving responsibilities.
Instructors warn that abandoning online education to prevent cheating would disproportionately harm marginalized student populations. Even in physical classrooms, the transition back to high-stakes, proctored exams creates new pedagogical problems.
Oral exams, while effective, are resource-heavy and often impractical given high student-to-faculty ratios. Furthermore, standardized written exams are often preferred for their ability to minimize subjective bias in grading compared to more complex, creative assignments.
Writing assignments, once a cornerstone of higher education, are increasingly being cut from syllabi. One instructor noted they previously required students in a natural disasters course to write a screenplay for a Hollywood movie, which forced students to apply complex physical concepts within a creative framework. That assignment is no longer viable because LLMs can generate such content instantly, removing the incentive for students to engage with the material.
As the barrier between student effort and automated output vanishes, the fundamental structure of the modern classroom faces an uncertain future. Faculty members continue to search for evaluation methods that measure genuine understanding without relying on outdated or exclusionary testing models.
