Artificial intelligence (AI) is making university degree assessments "unfit for purpose," threatening to undermine academic integrity and the credibility of qualifications, according to a report from the UK's Office for Students (OfS).
The universities regulator has warned that AI poses a "generational risk" to higher education, as lecturers report students are completing coursework assessments "in minutes" with AI tools and receiving similar grades to their peers who do the work manually.
The OfS's new guidance highlights that many degrees rely heavily on unsupervised coursework, which can now be completed by AI without genuine student engagement and is difficult to detect. Students told researchers that peers admit to using AI like ChatGPT for entire essays, often receiving good grades and then repeating the process.
"It is incredibly disconcerting when I plan and write an essay completely by myself, and get the exact same grade as someone who used AI to write it," one student reported. Another added, "I've had people admit, peer-to-peer: 'I wrote this whole essay with ChatGPT.' Then they don't get caught."
A previous study found that 94% of students are using generative AI in their assessed university work. The OfS report, compiled with Advance HE, noted that the use of AI in unsupervised assessments "cannot be reliably detected" and that resulting scores may not reflect actual knowledge and skills.
To combat this, the OfS is calling for a greater emphasis on supervised, traditional assessments such as closed-book exams, presentations, and practical or performance-based evaluations. These methods, the report suggests, can provide a "high degree of confidence" in a student's learning.
In response, a Department for Education spokesperson stated that universities are responsible for preventing academic misconduct, including the misuse of AI. The regulator will intervene if institutions fail to deliver effective and valid assessments. A Universities UK spokesperson acknowledged that many institutions are adapting assessments, incorporating more in-person exams, presentations, and process-based evaluations, while also exploring ways to integrate AI in controlled environments. They emphasized that assessment reviews must be ongoing to keep pace with evolving AI technology and maintain academic rigor and fairness.