Only 10% of students always verify information from generative AI

A study carried out by the Escola Pia de Catalunya (EPC) and the Universitat Oberta de Catalunya (UOC) on the use of generative artificial intelligence (GenAI) at schools concludes that the main weakness of its use in education is not checking the information contained in its responses. Only 10% of students "always" verify the answers GenAI gives them. 23.8% of the schoolchildren said they "often" check them, while more than half said "sometimes" (32.9%) or "rarely" (21.7%), and 11.5% said they have never verified any responses.

The report, the first of its kind to be carried out in Europe, includes indicators of trends and behaviour patterns in relation to GenAI based on a large sample of 3,700 secondary school students at 17 Catalan charter schools. The study reveals that knowledge of GenAI among students is unequal: 34.8% said they were unfamiliar or very unfamiliar with this technology, while 46.6% said they were very familiar with it.

Moreover, it found that the main way of learning about GenAI is self-learning. According to Marta López Costa, a member of the team leading the study and one of the coordinators of the UOC's Education Research Group (GREDU) and professor at the Faculty of Psychology and Education Sciences, "schools play a significant role, but it is still uneven". The conclusions of the analysis show that students mainly use GenAI to generate written content. More advanced uses, including image, video, code and simulation generation or prompt engineering are less common.

The authors also found that the technology now forms part of everyday academic life, considering that 54.9% of the students said they use electronic devices to study for one to three hours every day, and 27.1% for more than three hours.

The role of the school: from tolerance to competency-based use

In light of the results of the study —among whom are also professor Antoni Badia and professor Lorena Becerril, researchers at the Research Seminar on GenAI and training, teaching, and learning (SINTE-GenAI) and faculty members in the same department as López—, López believes that if this technology is to be successfully implemented in schools, the priority is to "build a pedagogical culture for the use of AI". In this regard, she said that "the first step is for schools to take on an active role," given that the report shows that students already use GenAI, but they have taught themselves how. "Schools must move from tolerance or an ad-hoc response to an explicit, progressive and shared strategy," she said. "We need to move from intuitive use to a critical and systematized academic digital competency."

Ramon Puig, the driver of the initiative and a member of the Escola Pia de Catalunya's AI Committee, said that the age factor should be part of the equation. "The data show that schools cannot respond in the same way at all stages: in compulsory secondary education AI literacy is necessary, while authorship and the ability to defend the process followed should be required of upper secondary education students."

For López, teaching students how to check the information they've obtained from GenAI is the second challenge facing schools. "This should be a fundamental skill: verifying responses, finding reliable sources, detecting errors, understanding biases and not confusing textual fluency with veracity," said the researcher attached to the UOC-FuturEd research centre. In fact, the study generated a series of recommendations for schools, emphasizing the need for checks to become second nature when using GenAI for schoolwork. "Moreover, students need to know when GenAI is a legitimate tool, when its use should be limited and which part of the process should be their own work," said López. "This directly affects assessment: it's not enough to ask for an end product, you have to ask for evidence of the process, justification of decisions and traceability of the use of AI," she stressed.

For Puig, the key question is no longer whether students use AI, but how they use it. "As they move from stage to stage, this technology takes on a larger role in their academic work, increasing the risk of replacing essential parts of the learning process.
 

The importance of training teachers

Teacher training is also an important factor according to López: "If teachers don't share guidelines, students will receive contradictory messages: something will be allowed in one course, forbidden in another, and not even mentioned in a third." Consequently, she believes that "the key lies in teachers and institutions taking on a leadership role". In her opinion, GenAI is above all associated with producing texts, but she thinks that it could be used for much more: comparing explanations, generating questions, simulating roles, checking arguments, adapting materials, creating examples, helping to plan and supporting comprehension. "The aim is to prevent it being reduced to writing essays more quickly," she said. In a similar vein, Puig said that "the major pedagogical challenge is for teachers to become the reference point for students in the digital environment: not as 'AI police' but as a guide to verify, interpret and preserve authorship by students."

In this new context, should using GenAI become a basic skill learned at school in the same way as reading, writing and arithmetic? López thinks it should, but with a caveat: "It shouldn't be a question of simply knowing how to use a specific tool, but knowing how to live, learn, work and critically think in an environment in which AI will form part of a wide variety of social and professional practices. Young people must learn to ask good questions, interpret answers, detect errors, understand limitations, protect data, recognize biases, correctly attribute the use of this technology and decide when it is wiser not to use it." She explained that there are international digital competence frameworks, such as DigComp 3.0, that have already incorporated it, and concluded that GenAI "does not replace traditional basic skills: it makes them even more necessary. To use AI properly, you have to know how to write properly, reason properly, have prior knowledge and know how to verify.
 

Ten recommendations from the experts on the use of GenAI in schools

The team that participated in the study has drawn up a series of pedagogical recommendations to move from intuitive use to a critical and systematized academic digital competency based on sound pedagogical principles and teaching support. These recommendations can be summarized as follows:

1. Draw up a school protocol or set of regulations setting forth shared guidelines on what GenAI is, how it works and what its limitations are.
2. Make the school the main place for critical learning about GenAI.
3. Explicitly teach how to verify information generated by AI. Comparing sources must be an intrinsic part, not an optional recommendation.
4. Work on prompting as an academic skill. Make formulating, specifying, reformulating and adjusting prompts a learning objective.
5. Define clear guidelines on permitted, limited and unacceptable uses. Students need to know what it considered help, what is delegation and what is plagiarism.
6. Integrate ethics: privacy, biases, authorship and responsibility. AI is not just a technical tool, it involves moral and social decisions.
7. Use GenAI as a complementary resource, not as a substitute. It must enrich learning processes, not replace materials, interaction with teachers or cognitive efforts.
8. Diversify tools and uses beyond ChatGPT and text generation. Learners need to compare platforms, explore different functions and avoid a limited vision of GenAI.
9. Adapt the use of GenAI to each subject and learning situation. Its application will differ in disciplines such as languages, social sciences, mathematics, art and physical education.
10. Provide more support to students with less self-efficacy or digital know-how. With good support, GenAI can level the playing field, but leaving it solely to individual initiative may widen the inequality gap.
     

About the Escola Pia de Catalunya (EPC)

The EPC is a socially committed educational institution that has more than 20 schools throughout Catalonia and an online study centre (Escola Pia Online). It currently teaches 20,000 students at all educational stages: from pre-primary education through to upper secondary education and vocational training. The institution has formed an Interdisciplinary Committee on AI, which has made decisions on the use of GenAI in schools or data protection, among other initiatives.

Related report

Badia, A., Becerril, L., López-Costa, M. (2026). L’ús de la intel·ligència artificial generativa (IAG) en els centres educatius de les Escoles. https://hdl.handle.net/10609/155389

This research is aligned with the UOC's research mission Education of the future and with UN Sustainable Development Goal 4, Quality Education. 

Transformative, impactful research

At the UOC, we see research as a strategic tool to advance towards a future society that is more critical, responsible and nonconformist. With this vision, we conduct applied research that's interdisciplinary and linked to the most important social, technological and educational challenges.  

The UOC’s over 500 researchers and more than 50 research groups are working in five research centres focusing on five missions: lifelong learning; ethical and human-centred technology; digital transition and sustainability; culture for a critical society, and digital health and planetary well-being.

The university's Hubbik platform fosters knowledge transfer and entrepreneurship in the UOC community.

More information: www.uoc.edu/en/research