94 per cent of students use AI. More than one in two (53 per cent) have also used it as a ‘digital confidant’

AI is already part of pupils’ lives. At least, that is what emerges from the survey ‘Me and Artificial Intelligence’, conducted by Indire and the Festival of Educational Innovation, which was presented today, 14 September, in the Senate, which involved 5,342 pupils from lower and upper secondary schools across the country. This provides a very realistic snapshot of the situation, and one that is worth understanding, not least because, as the organisers themselves point out, ‘in terms of scale, this is one of the most extensive surveys carried out to date in Europe on the use of artificial intelligence by students’.

The data reveal a phenomenon that has now become an integral part of everyday life. Almost all students – 94 per cent, to be precise – use artificial intelligence tools, and 62 per cent do so at least once a week. 84 per cent also use them for study purposes, ranging from Italian to maths, from writing to completing exercises. In some schools, 70–80 per cent of pupils report using digital technology – AI and other tools – on an almost daily basis.

Around 44 per cent fear that AI could lead them to think less for themselves, whilst 26 per cent cite the loss of critical thinking as one of the main risks for the future. Just over half of young people – 54 per cent, to be precise – want artificial intelligence to remain under human control; and 53 per cent have used it at least once to let off steam or deal with a personal problem, with a particularly high proportion among girls and younger students. 77 per cent use the free versions of AI services.

One of the most significant findings to emerge from the research is what the researchers have termed the ‘trust paradox’: those who trust AI blindly and, at the other end of the spectrum, those who say they do not trust it at all tend to verify the information they receive less frequently. It is, on the other hand, the students who adopt a middle ground – one of moderate trust – who carry out checks and verifications most frequently. Familiarity with the tool, therefore, does not automatically equate to the ability to critically evaluate its responses.

In presenting the results in detail, Samuele Calzone, a senior researcher at Indire, highlighted how the research reveals a significant gap between theoretical knowledge of the risks and actual behaviour. ‘Around a quarter of students are extremely aware of the risks associated with a loss of critical thinking and are fully aware of the limitations, errors and “hallucinations” of artificial intelligence,’ explained Calzone. ‘Yet this awareness does not automatically lead to greater scrutiny of sources. Knowing what one ought to do does not necessarily mean doing it.’