Daily Maverick columnist cites problems with university assessment due to AI
In South Africa, Daily Maverick columnist Steven Boykey Sidley said that generative artificial intelligence is making it harder for universities to verify whether students completed assessed assignments independently. In his view, traditional assessment methods are becoming increasingly less able to confirm an individual student's knowledge and raise questions about what exactly a degree certifies.
As an example, the author describes a situation in which a student submits a well-written essay before the deadline, while a lecturer suspects the use of a chatbot but cannot prove it. Sidley believes that when AI can write essays, generate code and solve problems, universities must redefine whether a degree certifies mastery of a subject or the ability to use AI tools.
Students' use of AI
According to data from a survey by the British think tank Hepi, published in February 2025 and cited in the column, 92% of undergraduate students used generative AI in some form, while 88% used it in assessed work. Another study, mentioned by the author and dated February 2026, found that 86% of surveyed students used AI, but only 23% considered themselves adequately prepared to use it. Most respondents reported insufficient guidance from educational institutions.
The columnist also notes that university policies on AI use differ significantly. In his assessment, academic research and regulatory decisions are failing to keep pace with the rate of development of artificial intelligence models.
More current news is available on the UA.News Telegram channel Telegram.
Limitations of detectors
According to independent testing of 14 tools for detecting AI-generated texts, cited by Sidley, none exceeded 80% accuracy. The University of Cape Town switched off its detector in October 2025. According to the author, the University of Waterloo, Curtin University, Vanderbilt University, Washington State University and other institutions made similar decisions.
Australian Catholic University generated nearly 6,000 accusations in one year, a quarter of which were not confirmed after review, after which the institution abandoned the tool. In its guidance, the UK's Quality Assurance Agency for Higher Education acknowledged that some or all assessments may no longer be confidently attributed to a particular student.
Among the alternatives, universities are testing oral examinations and multiple-choice tests. Sidley notes that oral examinations require significant time and resources, while tests can assess only certain types of knowledge. He also suggests that part of the verification of professional competencies may shift to licensing examinations, internships and interviews in relevant fields.
