Universities scramble to rethink exams as 92% of UK students now rely on AI for coursework

Why universities need to radically rethink exams in the age of AI

Since the launch of the chatbot ChatGPT in late 2022, educators have been grappling with how to harness artificial intelligence to enhance learning while minimizing risks to educational outcomes and the fairness of assessments. AI use among students is now the norm. In February, a survey of more than 1,000 full-time UK undergraduates found that 92% use AI in some form, up from 66% in 2024. And 88% of students reported relying on generative AI (a form of AI that can create text, images and code from vast data sets) to support their academic coursework, compared with 53% in 2024. As AI continues to outperform humans in basic tasks such as reading comprehension and computer programming, concerns have been mounting about its impact on learning and academic integrity. For example, the value of conventional essays and other written assessments is increasingly in doubt, given that AI can now produce writing that often surpasses the quality of most student work. Other concerns include an over-reliance on chatbots leading to superficial learning, reduced opportunities for self-reflection and a loss of student agency, with students becoming passive users of technology rather than active learners. Universities have responded by using tools to try to detect student use of generative AI. But these have proven to be unreliable. This has led to short-term fixes such as 'stress-testing' written assessments and replacing them with oral examinations, handwritten tests or reflective formats (portfolios and journals; see go.nature.com/43btcxf), as well as clearer guidelines on when AI can and cannot be used. Although these measures help, their effectiveness is limited. Instead, a fundamental rethink of learning and assessment is needed. Here, we highlight three promising approaches to examination that adapt existing methods -- such as conversation-based assessments -- to the AI era. These strategies aim to foster genuine intellectual development while ensuring that evaluations accurately reflect students' understanding and skills. One of the cornerstones of modern education is that 'writing is thinking'. Writing is a non-linear process that requires authentic engagement, critical thinking and problem-solving. All of these activities stimulate human intellectual development. When AI assists with or generates student texts, however, it becomes nearly impossible to know how much of the final work reflects the student's own understanding and critical thinking (see go.nature.com/47tjv93). This uncertainty undermines the use of writing as evidence of learning. Having a student and teacher follow a structured conversation is one way to enable critical thinking. For example, the Socratic questioning method is a form of disciplined inquiry that helps students to work through complex ideas, question their assumptions and judge the validity of information. In ancient Greece, the value placed on intellectual dialogue was so great that some philosophers at the time expressed concern that an over-reliance on writing could weaken human memory (see go.nature.com/43grxsp). A contemporary version of the discourse-based approach used in ancient Greece, known as conversation-based assessment, has been used for several decades in primary, secondary and university education settings. For example, AutoTutor, developed at the University of Memphis in Tennessee, has been used to teach subjects such as Newtonian physics while improving skills in computer literacy and critical thinking. It engages students in natural-language conversations and uses computational techniques to gauge their understanding -- analysing factors such as accuracy, choice of words and the time taken for a response. However, such systems usually have limited conversational capabilities and still rely mostly on simple text analyses and the detection of specific words and expressions. This is where the integration of AI can be a game changer. AI can sustain open-ended, context-sensitive dialogue in a much more realistic manner than current conversation-based assessment methods can. AI tools can ask students follow-up questions, provide tailored hints and adapt to a student's level of knowledge in real time, providing flexible and personalized learning support. And its questioning can range more widely than that of conventional conversational assessment systems, which are usually specialized to a particular domain. The crucial opportunity of AI is not merely to automate question-answering, but to enable students to learn through conversation with AI systems and to use that dialogue as a form of assessment, making it a dynamic and personalized process. Challenges remain. First, AI systems will need to guide conversations in a balanced way -- encouraging students to ask questions, explore topics that interest them and take an active role in their learning. At the same time, the dialogue must be structured enough for the AI system to gather meaningful evidence of the student's understanding, such as how they reason through a problem, explain a concept or apply knowledge in context. Achieving this balance between open-ended exploration and measurable assessment remains a major research challenge. Miscommunication is another concern -- AI systems might misunderstand a student's intent or provide inaccurate or misleading information. When this happens, students can struggle to identify the sources of their errors. The highly personalized and open-ended nature of AI-based learning and assessment would also make standardization difficult. So, there would still be a place for conventional assessments, especially in the university admissions process, in which consistency and fairness across large student populations is a priority. One crucial issue with many proposed responses to widespread student adoption of AI is that, although they attempt to safeguard academic integrity, they continue to operate within a high-stakes exam model. Even if the exam is reframed as a conversation, students remain aware that the outcome carries a lot of weight. Students find high-stakes exams stressful and might underperform or be tempted to cheat. The key challenge, then, is to reduce the need for high-stakes exams in an AI-fuelled era in which cheating might become easier. Continuous assessment can be an effective alternative. Replacing end-of-term exams with a series of interconnected assessments that build a comprehensive picture of student learning is urgently required in many academic fields. Continuous assessment is well established in medical education. For example, during clinical rotations, medical students are continually assessed by supervisors who observe their clinical reasoning, teamwork skills and communication with patients. These observations, combined with written reflections and peer evaluations, create a holistic picture of the student's competence over time. However, such models remain rare in other disciplines, mainly because of the increased workload they place on educators. The growing availability of AI-based systems makes continuous assessment more feasible. Conversations between students and an AI tool can be seen not as one-off exchanges, but as part of an ongoing learning process in which multiple low-stakes interactions gradually build a rich picture of student progress. The main challenge lies in ensuring that AI systems can track and analyse this learning progression effectively. Current general-purpose tools such as ChatGPT, Gemini and Copilot are not designed for this purpose -- they do not analyse students' responses over time to identify growth or persistent misconceptions. To truly support continuous assessment, there is a pressing need for learning-oriented AI platforms that can capture longitudinal data on student performance, provide meaningful insights into learning trajectories and integrate seamlessly into the design of courses and programmes. The resulting qualitative data would offer a richer, more holistic view of student progress than current approaches do, because such data would take into account a student's particular state of knowledge and developmental needs. Frequent, low-stakes tasks also reduce stress and the temptation to cheat, encouraging authentic engagement. Research shows that low-stakes assessments, in general, decrease academic dishonesty and student anxiety. Universities might also need to shift their focus towards developing students' higher-order and interpersonal skills -- such as creativity, collaboration and empathy. These are areas in which human strengths remain distinct and valuable. Instead of conventional written exams, students could, for instance, work together to design a sustainability plan for their university, create a prototype for a product or analyse a real-world policy problem. Unlike today's group projects, students could be encouraged to use AI tools throughout the process -- to brainstorm ideas, analyse data or structure their presentations -- with the AI platform serving as a university-authorized partner in learning. AI has been shown to support students during collaborative brainstorming sessions. However, assessing whether students are acquiring the necessary higher-order skills remains challenging. There is no consistent framework for evaluating creativity, collaboration or critical thinking across disciplines, which makes it difficult to compare results or validate the effectiveness of AI as a genuine educational aid. Although there is great potential for AI to improve learning in a meaningful way, many important issues remain to be addressed. First, educators and students need to strengthen their AI literacy. Educators must be given proper support -- through professional development, training and dedicated time -- to experiment with AI tools and integrate them into their teaching. Students need guidance on how to use AI safely and effectively, so that they understand its potential, limitations and ethical implications. Universities also need to develop clear policies on AI use that promote responsible and effective engagement. This will require rethinking what constitutes originality, critical thinking and creativity in student work. Such policies must also take into account the growing diversity of AI applications -- from general conversational systems such as ChatGPT and Copilot to specialized tools for research (such as Scite.AI), writing support (Grammarly, for instance) and image generation (such as Adobe Firefly). Institutions must also foster a culture that supports thoughtful AI integration. Rethinking learning and assessment on this scale will require a shift in organizational attitudes towards AI. The successful adoption of technologies depends heavily on how well they align with users' values and beliefs. In universities, both students' and educators' perceptions of AI shape how readily it is accepted and used. Investing in AI literacy across academia is therefore not just a technical necessity, but a cultural one. It is essential to ensuring that AI enhances, rather than undermines, learning and assessment.

Students embrace AI as schools tread carefully

At the Shenzhen College of International Education, a group of teenage students planning to attend universities abroad explain how generative artificial intelligence helps them with their work. Anita says the tool helps explain ideas she is studying while Lucienne uses it for research. Their enthusiasm lays bare a growing challenge for education at all levels: whether the use of AI by students and researchers should be restricted, grudgingly accepted or even actively encouraged. What is clear is that use of AI is growing quickly. A recent survey by the College Board -- a non-profit which organises exams in the US -- suggested the share of high school students using generative AI for schoolwork increased from 79 per cent in January this year to 84 per cent in May. In a study conducted by Oxford University Press in August, four-fifths of students aged 13 to 18 across the UK said they used AI tools in their schoolwork. Research for the UK's Higher Education Policy Institute showed the proportion of students using generative AI for assessments jumped from 53 per cent last year to 88 per cent in 2025. But for all the evidence of uptake, it is less clear precisely how generative AI is used and the implications of that for learning. Many teachers and students raise concerns about cheating, sparking some institutions to try to ban the technology or to get students to promise not to use it. John King, the chancellor of the State University of New York, for instance, says in some instances his colleagues are reverting to handwritten exams or oral tests. A study by AI developer Anthropic earlier this year examined the prompts students entered into its tool Claude. Many requests were "transactional" -- asking directly for answers to assignments, rather than "conversational" prompts to help them explore and understand concepts. On top of that, a large share of the queries required creativity or critical thinking from Claude, sparking concerns over whether students were outsourcing deeper intellectual tasks to computers instead of reflecting on the material themselves. Recognising such concerns, OpenAI, creator of ChatGPT, has this year unveiled a "study mode" feature that helps students work through questions step by step, designed to encourage their critical thinking rather than simply produce answers. Some see value in generative AI systems guiding students towards better understanding, although Aly Murray, founder of UPchieve, a company which matches volunteers with lower income students across the US, says: "AI tutoring is not a replacement for a human tutor and may never be . . . Simply knowing that the tutor is a human encourages students to learn more," she adds. "It's more motivating." Others are even more critical. A paper published this year by authors including Dirk Lindebaum at the University of Bath School of Management warns the use of large language models turns researchers in education from creators into simple consumers of knowledge, verifying information that is constrained by technology and defined by big tech companies. Higher education was first in denial, and is now integrating it as a technology. The third phase will be to integrate it into our curricula. Yet many schools and universities are starting to recognise exposure to AI is inevitable once students enter the workplace, so they should be provided with access, training and guidelines to best prepare them. Failure to do so risks putting some of them at a disadvantage in the job market. Approaches used by the institutions include asking students to explicitly disclose their prompts and to critique the machine-generated answers. Many are now offering courses focused on AI. Vallabh Sambamurthy, dean of the Wisconsin School of Business, says he has hired a series of younger specialist AI professors to help with teaching. "At least 15 per cent of our undergraduate business core courses should focus on AI topics," he says. Arizona State University, meanwhile, has struck a deal to roll out FYI.AI -- a tool backed by pop star will.i.am -- to its students. University president Michael Crow stresses the importance of personalisation, so that individual students can control the technology and hone it to their own needs. Joseph Aoun, the president of Northeastern University and a pioneer in thinking on AI whose book Robot Proof was first published in 2017, says: "Higher education was first in denial, and is now integrating it as a technology. The third phase will be to integrate it into our curricula." He believes higher education will survive by using AI as a tool while emphasising human skills that computers cannot replace such as entrepreneurship, creativity and teamwork. Hence his institution requires experiential, project-based learning and work placements -- aspects appealing to recruiters that a computer-written assignment cannot fulfil. At Shenzhen College, Arnold, another student, is already putting AI use into practice. He says the technology has become essential for his work: "I use it to understand concepts and translate ideas into English for my A-level exams and university applications." Costis Maglaras, dean of Columbia Business School, is relaxed about how this generation of students will cope. "I don't have concerns over the next decade," he says. "The real question is how things will be in 20 years for the undergraduates who have not yet been born. That's a wide open question."