New Coursera report: AI use by UK students doubles year-on-year, as educator confidence slips

New research reveals a sharp rise in AI use among university students in the United Kingdom (UK), while educators report declining confidence around both broader AI governance in higher education, and their own ability to navigate AI in the classroom.

• Almost half of UK students’ study tasks completed using AI: double last year’s amount and the highest of all countries surveyed by Coursera
• 4 in 5 students in the UK say that their grades have improved since they started using AI 
• Just 30% of UK universities have a formal policy on AI usage, still the highest of any surveyed country
• Only a quarter of educators are confident they can detect AI-generated work (down from 42% last year)

Coursera’s AI in Higher Education Report, released today, surveyed over 4,200 students and educators from universities in the UK, United States (US), Mexico, India and Saudi Arabia, and found that UK students are now using AI to complete nearly half (48%) of their studies (versus 24% in 2024). This is a higher proportion than the global average of 44% and that of US students, who are using it for 38% of their study tasks.

This increase in usage is delivering clear academic benefits, with 4 in 5 (80%) UK students saying their grades have improved since they began to use AI, with 43% saying they have significantly improved. This marks a substantial jump from last year (2024), with just over half (52%) reporting improved grades (and 8% identifying substantial improvement). 

AI adoption is also increasing among educators, with 68% now using AI often or always in their work. The proportion of UK educators who always use AI has risen from 21% in 2024 to 33%, slightly above the 30% global average.

Student confidence grows; educators express increasing AI anxieties 

• The majority (85%) of UK university students believe AI is having a positive impact on higher education — up from 67% in 2024. In contrast, educator sentiment in the UK is moving in the opposite direction: 
• The proportion of educators who believe AI is having a positive impact on higher education has fallen from 85% to 69% 
• In 2024, just over a fifth (21%) of educators did not think AI was having, or could have, a negative impact on higher education, but by 2025, that figure has fallen dramatically to just 3%

Fewer UK educators are feeling prepared to navigate the pedagogical and professional challenges presented by AI:

• They are less likely to believe they and their peers have the right skills to use AI to their advantage (27% vs 37%)
• They feel less confident that they can detect AI generated content (26% vs 42%)
• This uncertainty is felt internationally, with just 22% of educators in the US believing they can reliably detect AI-generated content

UK universities make progress on AI readiness and governance

Despite confidence gaps, the research finds that UK educators feel less unprepared to navigate AI than those of any other country. 46% of UK educators and students believe their system is unprepared for AI, down from 55% last year, and lower than  the 56% global average (US: 50%). 

This confidence is backed by stronger governance than other countries. While less than a third (30%) of UK universities have a formal AI policy, this is more than the 26% global average and just 20% of universities in the US.

Concerns around the effect of AI on academic credibility are also easing in the UK. Just over half (55%) believe unregulated AI could undermine degrees, the lowest level of concern among countries surveyed, and down sharply from 70% last year.

Marni Baker Stein, Chief Content Officer at Coursera, said: “AI is now ubiquitous throughout higher education. Students are uniformly embracing the technology to support their learning, while educators are increasingly adopting it to drive productivity and a more personalised pedagogy. However, as AI models become more sophisticated and evolve at pace, many educators are finding it harder to keep up with the practical and professional implications — contributing to a decline in confidence around skills and AI detection.

While the UK is ahead of many global peers, with stronger perceptions of institutional readiness and clearer approaches to AI governance, continued progress will demand a focus on training and professional development to build faculty confidence and capability. Coursera remains committed to supporting learners and educators to use AI responsibly and effectively.  We will continue collaborating with our institutional partners to ensure that technological progress continues to serve equitable, world-class education.”

Caroline Williams, Executive Director, Oxford Saïd Online, commented: “At Oxford Saïd, we see AI as a learning companion, not a content authority or a source of answers. Drawing on evidence-based learning design, we’re building online courses for 2026 where AI helps learners understand ideas, think critically, and create things together, acting like a thought partner. Central to this approach is a responsible and transparent human-AI relationship, one that deliberately evolves from structured support toward learner agency, critical engagement, and purposeful use, ensuring that cognitive responsibility remains with the learner. Ultimately, our aim is not to teach learners to rely on AI, but to learn with it thoughtfully, critically, and with intellectual ownership.”

Methodology 

This research was commissioned by Coursera and conducted by Censuswide, with a sample of 4,261 educators in universities (professors, lecturers, and seminar/discussion leaders) and university students (18+) in the United Kingdom (UK), United States of America (US), India, Mexico, and Saudi Arabia. The data was collected between October 15–23, 2025 using an online survey.

In 2024, Censuswide conducted UK research with a sample of 1,000 university students (18-24) and 501 educators in universities (Professors, lecturers, seminar/discussion leaders). The data was collected between 04.10.24 – 14.10.24. 

Censuswide abides by and employs members of the Market Research Society and follows the MRS code of conduct and ESOMAR principles. Censuswide is also a member of the British Polling Council.