Artificial Intelligence in art classrooms

Between experimentation, bias and new educational responsibilities, Artificial Intelligence enters the teaching of artistic practice as an ambivalent tool: a promise of exploration or a risk of homogenisation? Starting from the state of the art of research on the use of AI in this field and from the reflections of the communicators of CoMeta APS, engaged in digital inclusion and technological creativity, here is what needs to be known.

The smell of animal glue, marble dust settling on the hands, the sharp sound of the chisel striking stone. In Renaissance workshops, learning happened like this: by watching, making mistakes and repeating. Artistic training did not pass through manuals, but through the body, the slow time of learning and proximity to the master. It is said that Michelangelo, still very young, learned more by observing the hands of other sculptors than by listening to their words: art was transmitted as a gesture, even before it was an idea. Even today, in classrooms and studios, education in artistic practice retains this concrete and sensitive dimension: it is made of materials, attempts, errors and choices that arise from the friction between mind and hand. Into this space a new interlocutor has arrived: Artificial Intelligence. How can it be integrated? Starting from the state of research on the use of AI in art classrooms and from the reflections of the communicators of CoMeta APS, here is a practical compass on what needs to be known to integrate it into art education.

The filters of creativity: cognitive bias and algorithmic bias

Can Artificial Intelligence really support students and teachers in art classrooms without becoming a shortcut that simplifies the creative process? Before answering, it is useful to look at the heart of art itself: every work is born from a process of analysis and synthesis, in which critical thinking, culture and the ability to observe the world play a central role. When an artist chooses colours, materials, forms or tools, they always do so through conscious and unconscious filters: the creative act is never random, it is a dialogue between mind, hand and expressive medium. When thought translates into form, the biases that guide its direction often emerge silently. Cognitive biases are automatic patterns of thought that the human brain uses to interpret the world quickly. They are filters that help us make fast decisions, but they can also distort reality: confirmation bias, for example, leads us to favour information that confirms what we already believe to be true; availability bias makes us overestimate events that are easily recalled from memory; cultural bias conditions our aesthetic taste and our perception of forms and colours based on the cultural context in which we live¹. From this perspective, teaching art and educating in artistic practice means accompanying students in building a gaze that can recognise this complexity and take the time to observe, to question the reasons behind certain formal and conceptual choices².

Moreover, when the use of AI enters art education, recognising the existence of these cultural and cognitive filters helps to remove a very common alibi: that of attributing every distortion or error to technology. Neither human creativity nor, even less so, the operation of Artificial Intelligence act in a neutral space, but process and amplify structures that are already present. Artificial Intelligence presents algorithmic biases that derive from the data and mathematical models that feed it. An algorithm trained predominantly on Western images will tend to produce results aligned with that cultural context, replicating already consolidated patterns³.

We recommend reading

21 June 2025

A cultural object called dataset

When archives and curatorship meet AI

Artificial Intelligence Data Art Photography

For this reason, it is only by relating these two levels of filtering, human and computational, that it becomes possible to critically read what AI produces. Designer and academic Hamid Amini underlines that the risk of simplification exists only if the use of technology is disconnected from the educational context and from the guidance of the teacher⁴. When, instead, AI becomes an integrated tool in the workshop, whose limits and potential are known, it takes on an exploratory function: it offers prompts that can be interpreted, modified and transformed into a manual or multimedia work. Students learn not to passively accept the output, but to read its recurrences and biases, to recognise why certain bodies, faces and styles return incessantly. The creative process remains fully in human hands, which must choose, judge and reflect on the decisions taken, on the biases at play, on the prompts sent and the content generated.

How to make it happen: multilayer workshops

Research conducted in 2024 by Sara Sáez-Velasco of the Faculty of Education at the University of Burgos, together with Mario Alaguero-Rodríguez, Vanesa Delgado-Benito and Sonia Rodríguez-Cano of the Faculty of Humanities, analysed the impact of generative AI in art education in upper secondary education, collecting opinions from both teachers and students⁵. The research was carried out on a sample of ten participants (five students and five teachers) involved in art education and visual communication pathways at the University of Burgos in Spain. The students were part of the Master’s programme in Communication and Multimedia Development, a university-level course oriented towards digital creativity, communication and media technologies, with skills in illustration and digital tools; the teachers belonged to the Department of Audiovisual Communication and Advertising, teaching technical and creative subjects such as illustration and image analysis. In this study, multilayer workshops were used, that is, a teaching methodology structured across several experimental levels, which involved distributing different roles and activities among participants to stimulate discussion, reflection and practical skills. The methodology included, among other things, experimental tests to generate visual content using different techniques and discussion groups focused on the perception and experience of the tools. Some participants interacted directly with generative tools, others transformed ideas into manual works, while others curated the visual narrative or documented the process⁶. AI was perceived as a creative support, capable of expanding expressive possibilities without replacing the decision-making autonomy of students. 85% of participants stated that AI expanded creative possibilities without limiting their artistic decisions, and one student commented at the end of the lesson:

«I can experiment more without losing control over the result», returning an image of the tool as an ally of artistic practice⁷.

Starting from this theoretical framework, it is possible to imagine many other educational workshops in which AI is used as a starting point for creation⁸. For example, a workshop could begin with a generative prompt to produce a digital landscape: some students interpret and manually redraw elements of the landscape, others produce short narrative texts to accompany the image, while others experiment with digital stylistic variations. At each stage, the class observes what the AI proposes, discusses recurring patterns and compares aesthetic choices.

Conscious visions of AI

Every output produced by an algorithm raises questions of cultural equity, copyright and environmental impact, even in the field of education in artistic practice. For this reason, the UNESCO report Artificial Intelligence and Culture underlines the importance of ensuring transparency, equitable access to tools and critical training⁹. In line with UNESCO’s recommendations, the European AI Act provides a regulatory framework that reinforces the principles of equity, transparency and responsibility, establishing clear criteria for the use of generative AI systems, such as the obligation to make artificially generated content recognisable, the traceability of processes and training data, the protection of copyright and the prevention of cultural bias, thus supporting an educational approach based on critical awareness and the promotion of cultural diversity¹⁰. For educators and teachers, this translates into developing analytical tools that not only reveal the play of biases but also create awareness of the other issues at stake, without ever neglecting the artistic gesture. These include frameworks for reading generated content, which invite students to question the artificial origin of texts and images, the context of production and communicative objectives, bias analysis sheets and practices of process traceability, documenting prompts, revisions and decisions made during interaction with AI, making explicit the active role of the human being in co-creation.

An example, from Italy this time, comes from CoMeta APS, which with the project Futuro Consapevole, funded by the European Union, is bringing awareness-raising pathways into schools, libraries and cultural centres on the role of digital technologies in contemporary culture. The project offers workshops, public events and informational meetings dedicated to themes such as Artificial Intelligence, immersive reality and digital identity, with the aim of educating young people and citizens in the responsible use of emerging technologies and stimulating a critical and conscious approach to their integration into everyday life. These programmes are structured through public debates, interactive workshops, presentations of real cases and open calls for artistic and cultural projects, encouraging especially those who are not fully familiar with these digital tools to engage with concepts such as bias, digital identity, ethical impact and the future of work. Events take place in various libraries and cultural spaces across Italy, involving students, educators and community members in open conversations about the risks and opportunities of digital technologies, rather than simple frontal lectures.

From their point of view, Artificial Intelligence should not be seen as a shortcut nor as a substitute for creativity, but as genuine support. Students, they explain, can use AI to suggest combinations, generate drafts or explore styles, but the critical value lies in the choices they decide to make: formulating and modifying prompts, discussing results, interpreting and transforming ideas into manual works or physical installations. AI becomes a tool for dialogue and comparison, stimulating collaboration between those working on digital aspects, those focusing on craftsmanship and those on narrative, without simplifying or reducing the creative process. As Mario Esposito states:

«It is essential that students and teachers do not limit themselves to using AI as an operational tool, but learn to question what it produces, carefully evaluating choices, limits and possible cultural exclusions».

 

1. “Cognitive bias”, defined as “systematic errors in the way individuals reason about the world due to subjective perception of reality” by Britannica, i.e. mental filters that help to interpret information quickly, but which can distort reality. https://www.britannica.com/science/cognitive-bias ↩︎
2. Busch, L., Malkin, A., & Belisle, J. (2025). Art in context: A multi-level analysis of art. Journal of Contextual Behavioral Science, 36, 100890. https://www.sciencedirect.com/science/article/pii/S2212144725000213. This study analyses art as a complex phenomenon in which creativity, cultural context, behaviour, and symbolic dimensions interact, showing how artworks arise from an integrated process of perception, interpretation and social construction. ↩︎
3. Chinta, S. V., Wang, Z., Yin, Z., Hoang, N., Gonzalez, M., Quy, T. L., & Zhang, W. (2024). FairAIED: Navigating fairness, bias, and ethics in educational AI applications. arXiv (Cornell University). https://arxiv.org/abs/2407.18745 ↩︎
4. Amini, H. (2025). AI in Art Education: Innovation, Ethics, and the Future of Creativity. Art And Design Review, 13(02), 115–129. https://www.scirp.org/journal/paperinformation?paperid=142478 ↩︎
5. Sáez-Velasco, S., Alaguero-Rodríguez, M., Delgado-Benito, V., & Rodríguez-Cano, S. (2024). Analysing the Impact of Generative AI in Arts Education: A Cross-Disciplinary Perspective of Educators and Students in Higher Education. Informatics, 11(2), 37. https://doi.org/10.3390/informatics11020037 ↩︎
6. Chinta, S. V., Wang, Z., Yin, Z., Hoang, N., Gonzalez, M., Quy, T. L., & Zhang, W. (2024b). FairAIED: Navigating fairness, bias, and ethics in educational AI applications. arXiv (Cornell University).https://arxiv.org/abs/2407.18745 ↩︎
7. Sáez-Velasco, S., Alaguero-Rodríguez, M., Delgado-Benito, V., & Rodríguez-Cano, S. (2024b). Analysing the Impact of Generative AI in Arts Education: A Cross-Disciplinary Perspective of Educators and Students in Higher Education. Informatics, 11(2), 37. https://doi.org/10.3390/informatics11020037 ↩︎
8. Vartiainen, H., Tedre, M., & Jormanainen, I. (2023). Co-creating digital art with generative AI in K-9 education: Socio-material insights. International Journal of Education Through Art, 19(3), 405–423. https://doi.org/10.1386/eta_00143_1 ↩︎
9. UNESCO Independent Expert Group (2025). Artificial Intelligence and Culture (CULTAI Report). United Nations Educational, Scientific and Cultural Organization. https://www.unesco.org/sites/default/files/medias/fichiers/2025/09/CULTAI_Report%20of%20the%20Independent%20Expert%20Group%20on%20Artificial%20Intelligence%20and%20Culture%20%28final%20online%20version%29%201.pdf ↩︎
10. European Parliament. (2023, December 9). Artificial Intelligence Act: deal on comprehensive rules for trustworthy AI. Comunicazione stampa. https://www.europarl.europa.eu/news/es/press-room/20231206IPR15699/artificial-intelligence-act-deal-on-comprehensive-rules-for-trustworthy-ai ↩︎