AI in universities: almost all of them provide training for students and lecturers, but only 25 per cent have a dedicated fund. The Luiss-Google model for rethinking courses

In a national context where almost all universities are training students and lecturers in artificial intelligence, only one in four has a dedicated budget. For the first time, a survey carried out by CRUI, the Conference of Rectors of Italian Universities, has taken stock of the situation in the country; the findings were presented yesterday in Rome, at Palazzo Brancaccio, during the third edition of the Higher Education Leader Series, an event organised by Google for Education. These key points formed the basis of the speech by Gianluigi Greco, CRUI’s ICT delegate and Vice-Chancellor of the University of Calabria, who noted that the academic system has been sharing best practices on generative models ‘for years’, ‘from teaching to research, from day-to-day management to student services’. In terms of adoption, digital literacy and governance policies, he added, the system “has some very advanced areas, but also areas that require support and strengthening”.

Looking at the figures, in fact, 93 per cent of universities have launched at least one AI literacy course. By 2026, there will be 978 such courses: 334 for the general public, 264 for lecturers, 225 for students and 155 for staff. Almost all of them consider AI to be fundamental to their objectives, and 65 per cent have adopted guidelines, whilst a further 33 per cent are currently discussing them. Resources, however, are not keeping pace: only 25 per cent have a dedicated fund. Of the 53 use cases surveyed, 87 per cent are concentrated in administration and teaching, with research accounting for just 13 per cent. The weak point is risk management: 62 per cent view it solely as a theoretical principle, 51 per cent have no procedures in place to detect misuse, 77 per cent of regulations make no mention of the data protection officer, and 82 per cent of guidelines do not provide for sanctions. Among lecturers, cultural resistance and a lack of incentives are significant factors; amongst students, the risk of cognitive delegation is a concern. According to Greco, we need to move beyond the phase of fragmented experimentation towards ‘systemic and structured’ governance. The role of the CRUI, he explained, is to ‘define common guidelines to support the transition’, as part of an innovation process ‘geared towards meeting people’s needs’. The crux of the matter is training graduates capable of using these tools critically. For Colin Marson, Director of Google for Education for the EMEA and JAPAC regions, the turning point lies in ensuring ‘that teachers retain a guiding role’, with the necessary skills to support students. However, we must not fall behind, because the market is moving fast: of the 31 million entry-level job adverts analysed by Inco in collaboration with Chance, 24 per cent require artificial intelligence skills.

According to Marco Berardinelli, head of Google for Education in Italia, our country is among the most advanced in Europe across the entire education system, from primary school teachers – who use AI mainly for visual materials – right through to sixth-form colleges. And universities are following suit: Gemini for Education is used by over a million students in more than half of the country’s universities.

Universities are gearing up, and there are a number of active initiatives across the country. Among the initiatives presented was Luiss, represented by Enzo Peruffo, Vice-Rector for Teaching, and Livia de Giovanni, Vice-Rector for Rankings, Accreditation and Quality. Following the agreement with Google in March, the university has made Gemini and Notebook available to students, lecturers and staff in an environment where data remains within the university. Over the last two weeks, Gemini has reached nearly a thousand active users per day, with over 76 per cent using it on at least two days. On Notebook, the most common activities are chatting with AI (22.7 per cent) and uploading documents (20.3 per cent). In addition, there is the Gem Wall – a library of study and teaching assistants – as well as training sessions for lecturers and staff, and grants for teaching innovation.

An AI Handbook sets out, course by course, how AI can be used. On this point, it should be made clear that integrating these tools is simpler in quantitative disciplines, where students were already using them for exercises, whilst it is more complex in subjects such as management. In such cases, the onus falls on lecturers, who are the driving force behind this change: as Peruffo emphasises, courses need to be almost entirely rethought, not just the final exam out of fear of cheating. And time is of the essence, because this is what students and businesses are demanding; they are looking for graduates capable of using these tools with a critical eye.