A peek into lecture theatres reveals that artificial intelligence (AI) is changing higher education. Lecturers, researchers and students are all equally benefiting from this progressive development – thanks to new options for making learning more efficient, individual and accessible.

Oscillations, fields, waves – the world of physics is fascinating, but also often difficult to grasp. Dr Paul Kuria Kamweru, Associate Professor of Physics at Chuka University in Kenya, relies on artificial intelligence to vividly convey abstract phenomena by means of visualisation. ‘Simulations make a decisive contribution to better comprehension, especially when it comes to a traditional topic like electromagnetic waves’, he says. Kamweru also uses the new technology for interactive presentations and exam questions: ‘I use it to generate assignments that cover all levels from basic knowledge to critical application and evaluation’, he says. This enables him to adapt the level to suit bachelor’s, master’s degree or PhD students. The tools also enable him to convert theoretical content into practical scenarios – and to add realistic data as necessary. Kamweru is convinced: ‘AI-based tools make my work more efficient, my teaching more vivid and learning more accessible to my students.’

Individual learning during studies: innovative teaching methods using AI

Dr Tosin Adewumi, postdoc at Luleå University of Technology in Sweden and an expert in machine learning and innate language processing, also sees huge potential for the use of AI in the education sector. ‘Having around 500 students in our mathematics courses makes it barely possible for our teaching staff to provide personal mentoring’, he explains. AI is instead used in a targeted manner to support individual learning and provide direct feedback. One example of this is his innovative teaching concept that combines a number of approaches: a ‘Socratic dialogue’ with purposeful interposed questions prompts independent thinking, whereas the ‘chain of thought’ thinking by which the AI explains its thinking facilitates understanding of more complex topics. Playful elements additionally provide variety, and the students obtain personalised feedback on their learning progress.

The reason many lecturers still hesitate to use AI

Why many lecturers worldwide nonetheless hesitate to incorporate AI in their teaching is revealed in a case study entitled ‘Generative AI and Teachers – For Us or Against Us?’, in which Tosin was involved. ‘The global uncertainty at higher education institutions is substantial’, he explains. ‘Students are though already using AI anyway, so I deem it expedient to teach them how to use it in a considered and responsible manner.’

AI in research: more efficient analyses and creative support

Because research too can benefit from AI as a powerful tool: for analysing large data volumes, detecting patterns, generating hypotheses or simulating more complex systems. Automated literature research is also useful: relevant publications can be more easily identified and organised using AI-based search and classification systems. So-called large language models (LLMs) are used to generate ideas or create documents. ‘We’re experiencing a massive transformation in the higher education sector due to the wide range of opportunities offered by AI’, says Adewumi.

New exam formats and risks due to AI misinformation

It also involves an alteration to exam formats: some higher education institutions are increasingly resorting to oral exams given their concerns regarding misuse of AI, such as plagiarism in written assignments. Adewumi believes there is an even greater risk associated with its use – misinformation due to so-called AI hallucinations. ‘This can however be significantly reduced by using technological models like retrieval-augmented generation systems, which include information from reliable sources’, he says. Other difficulties like bias in the training data, potentially in terms of gender or origin, require the user to apply critical awareness.

A critical approach to AI: training, ethics and challenges

Kamweru therefore emphasises the need for targeted training of lecturers and students along with clear guidelines on dealing with AI. ‘There is often uncertainty without binding guidelines – including in terms of data protection, ethics and permissible use’, he explains. ‘At the same time, the technical infrastructure isn’t always sufficiently available or fairly distributed, which can further compound the existing inequalities in the education system.’ He also warns about over-dependency: ‘AI used in an unconsidered manner can impair the capacity for independent thinking and problem solving.’ Albeit AI is felt to promote the democratisation of learning, since it offers access to high-quality materials for all learners and higher education institutions. ‘AI-based online courses can replace expensive textbooks and ensure flexible design of learning time’, says Kamweru.

Three central pillars for responsible use of AI in education

Not least for these reasons, he is calling for the sustainable integration of AI into higher education supported on three pillars: the consistent inclusion of ethical principles, the targeted promotion of critical handling of AI-generated content and the development of exam formats that make independent thinking the focal point. ‘AI could therefore be used responsibly and with positive effects on education.’