Universities in England face sanctions over AI assessment gaps
Universities in England could face financial penalties and additional regulatory requirements if they cannot reliably assess students’ knowledge as generative AI becomes commonplace. The Office for Students warns that unsupervised written assignments may no longer demonstrate genuine understanding, potentially undermining confidence in degrees. The regulator has not banned ChatGPT or announced automatic penalties for AI-generated student work; its focus is compliance with existing educational quality standards.
A Higher Education Policy Institute survey of 1,054 students found that 95% used AI for at least one activity and 94% used generative systems to help with graded assignments. Those uses included explanations, outlines and editing, rather than necessarily submitting generated work. Detection software offers no definitive solution: tools can miss generated content or incorrectly flag human writing. A Stanford study published in 2023 found that 61.22% of essays in a TOEFL sample written by non-native English speakers were incorrectly classified as AI-generated, although those findings cannot be generalized to all current tools. Turnitin cautions against using automated results as the sole basis for adverse decisions.
Recommended alternatives include supervised exams, oral questioning, practical demonstrations and reviewing how assignments develop. Universities can also assess students’ ability to critique AI-generated answers. Research on learning suggests that design matters: tools offering ready-made solutions can weaken independent performance, while specialized tutors that encourage reasoning can support learning.