Learning to use/through the use of ChatGPT: the AI literacies of Italian students.

As with other new technologies, children and young people have been eager early adopters of Generative AI tools since the release of ChatGPT in November 2022. Research worldwide has documented the most common uses of GenAI tools among this age group, as well as their expectations and fears for the future. For example, young people in the US tend to engage with textual Gen AI more than image- or video-based GenAI, and they do so mainly to get assistance with their homework or to escape boredom (Madden et al., 2024). However, research is still sparse and mainly descriptive, leaving room for unsubstantiated public discourses that either reproduce the commercial hype around the revolutionary opportunities opened by Gen AI; or replicate media panics around the harmful negative effects for children and young people, as if students’ cheating in school, adolescents’ mental health problems, and disinformation campaigns, were only determined by Gen AI. Moreover, the question of young people’s AI literacies is still underexplored. This presentation reports on interviews with 15 13-to-17-year-olds in Italy, as part of a qualitative comparative research across 13 European countries (Austria, Czech Republic, Estonia, Ireland, Italy, Latvia, Luxembourg, Norway, Poland, Portugal, Serbia, Spain, the UK). The in-depth interviews were conducted both in person and online, to a diverse sample of adolescents in terms of gender, age balance, school type, socioeconomic status, urban or rural area. We examine the AI literacies of young people, and prerequisites and barriers to developing AI knowledge and skills (including individual resources and vulnerabilities, and structural enablers and barriers, Helsper, 2021). Informed by a conceptualisation of digital literacy as consisting of knowledge and skills elements as well as functional and critical aspects (Smahel et al., 2024), we define AI literacy as the combination of critical skills and knowledge about how Gen AI works. The findings confirm that AI literacy consists of both functional and critical skills and knowledge, and is influenced by age and context. Adolescents use GenAI primarily for schoolwork, especially to help them with writing tasks or to support them in appropriating difficult concepts or summarising long texts. By contrast, they tend to distrust GenAI when it comes to solving math problems. Accordingly, many interviewees understand ChatGPT as a direct replacement for search engines. Adolescents learn to use GenAI by themselves, through a learning by doing process; or via informal learning practices - including word of mouth, or videos on Instagram or TikTok. Only occasionally they are introduced to positive uses of ChatGPT and other tools by teachers. Family discussions on GenAI, instead, remain sporadic and parents do not play a significant role in scaffolding adolescents’ skills. their skills and knowledge around GenAI is unevenly distributed. While most interviewees share a basic understanding of GenAI capabilities as stemming from internet data, they tend to lack foundational technical knowledge or awareness of underlying mechanisms, and ignore risks related to datafication or algorithmic bias. As a result of limited AI literacy, when presented with hypothetical scenarios, adolescents reproduce dystopian discourses, with reference to the risk of disinformation and a totally AI-fabricated reality, cognitive disempowerment (where increased reliance on Gen AI may undermine critical thinking and independent problem-solving skills), and automation replacing human roles and decision-making. In conclusion, our study highlights how adolescents’ engagement with GenAI is shaped by diverse and fragmented learning pathways, with home, school, and social media representing potential educational loci offering uneven but equally insufficient support. Our findings contribute to the debate on the need for comprehensive AI literacy education that promote critical, ethical, and informed engagement with GenAI.