Artificial intelligence (AI) adoption in higher learning, teaching and assessment is outpacing the policies, skills and ethical safeguards needed to govern it responsibly, according to research presented by Vaal University of Technology (VUT).
The study, presented at the Second International Conference on Climate Resilient, Smart and Sustainable Futures (ICCRSSF 2026) in Victoria Falls, Zimbabwe, examined how lecturers at VUT are integrating AI into teaching and assessment.
The conference, held from 12 to 14 August, brought together researchers, universities, policymakers, industry experts and development partners.
Titled: “Beyond online learning: Lecturers’ adoption of AI tools in teaching and assessment in a South African university of technology”, the study was presented by Naledi Kaeane, a lecturer and researcher in VUT's Department of Tourism and Integrated Communication.
One of the key concerns identified by the study is the lack of clear and consistent institutional guidance on acceptable AI use.
“Lecturers reported uncertainty around accountability, ethical boundaries and the use of AI-generated content, with institutional policy frameworks described as inconsistent or, in some cases, absent,” the study points out.
“This leaves individual lecturers having to navigate questions about whether and how AI can be used in teaching, assessment and student submissions. The issue also raises questions around the reliability of AI detection tools, student privacy, intellectual property and whether conventional assessment models remain appropriate in an environment where AI assistance is increasingly accessible.”
The findings come as universities globally grapple with how to respond to generative AI tools capable of producing essays, computer code, research summaries and other academic work.
South African universities are adopting increasingly cautious positions on AI detection tools. Some previously told ITWeb they have discontinued their use, while one has never adopted them, and many emphasise that AI-generated content detectors cannot be treated as conclusive evidence of academic misconduct.
The VUT research further found that the pace of adoption is creating uncertainty around how AI should be governed within higher education.
“Limited digital literacy is preventing some lecturers from using AI tools confidently and independently, while the growing use of AI-generated or AI-assisted student work is challenging established approaches to academic integrity and assessment.”
Not teaching replacement
The qualitative study found lecturers generally view AI as a tool that can improve teaching efficiency, enhance student feedback and support innovation in instructional design, rather than as a direct threat to their jobs.
“AI is not replacing lecturers, but rather, it is reshaping what it means to teach, assess and engage in higher education,” explained Kaeane. “Structured support, clear policy and ethical guidelines will determine how responsibly this shift unfolds.”
The study recommends that rather than banning AI, universities develop structured programmes to improve lecturers' digital literacy and confidence.
It also calls for institutional policies that define acceptable AI use, ethical guidelines to protect academic integrity and continued engagement between university leadership and lecturers on governance and accountability.
The researchers argue that universities should help lecturers understand both the capabilities and limitations of AI, while redesigning assessments where necessary.
“The objective should be to ensure technological innovation does not undermine the credibility of academic qualifications. Building lecturers' capacity to use emerging technologies responsibly will become increasingly important as AI becomes embedded in technology-supported classrooms,” the study notes.
According to Kaeane, the findings align with Sustainable Development Goal 4, which focuses on inclusive and equitable quality education and lifelong learning opportunities.
AI governance
The study highlights a broader challenge facing universities: AI adoption is not simply a technology issue, but increasingly one of institutional governance.
“While lecturers may see practical benefits from AI, universities need frameworks that establish where those benefits end and unacceptable use begins. This includes determining how AI-assisted work should be disclosed, how assessments should be designed and how responsibility should be allocated when AI-generated content is used in academic settings.”
VUT has already been grappling with the governance and ethical implications of AI in higher education.
In September 2025, the university hosted a workshop with the Technological Higher Education Network South Africa and the US Fulbright Scholarship Programme, where academics and experts examined issues including algorithmic bias, privacy, automation and the difficulty of distinguishing AI-generated student work from original research.
The workshop also highlighted the potential for AI to support personalised learning, innovative assessment design, online course delivery and student support. Participants argued that universities should move beyond punitive approaches and provide clearer guidance on when AI is prohibited, or can be responsibly used, while keeping students accountable for work submitted with AI assistance.
The main conclusion of the VUT workshop was not to ban AI, but to move towardsclear, supportive institutional guidance on responsible AI use in higher education. Participants agreed that universities should distinguish between assignments where AI is prohibited and those where it can be responsibly used, rather than relying primarily on punitive measures.
More broadly, the conclusion was that AI adoption needs ethical safeguards around fairness, transparency, reliability, accountability, privacy and bias, with technology serving education rather than undermining human creativity and critical thinking.
Share
