Towards AI literacy: A proposal of a framework based on the Episodes of Situated Learning

Synthesis: Panciroli, Allegra, Gentile & Rivoltella propose a framework for curricular AI literacy integration built on the Episode of Situated Learning (ESL) instrument. ESL is an active teaching device organized around three verbs — anticipate, produce, reflect — that operates at both the micro level (lesson planning) and macro level (curriculum construction). The authors situate AI literacy within the broader landscape of new literacies (media, information, data, AI), draw on European AI competence frameworks (EU Digital Education Action Plan, de la Higuera's five pillars, DigComp 2.2), and distinguish direct didactics of AI (learning how AI works) from indirect teaching of AI (learning to collaborate with AI). Both levels fall under an educational mode of "explainability" centered on social interaction and school-community participation.

AI literacy within the new literacies landscape

The paper situates AI education within the broader evolution of literacy concepts, following the New London Group's multiliteracy framework. It distinguishes at least four specific literacies with distinct competences: media literacy (comprehension of media messages), information literacy (finding, analyzing, using, sharing information), data literacy (understanding the reality of data and developing reflexive, critical interaction with it), and AI literacy (knowing, understanding, using, applying, evaluating, and creating AI). AI literacy thus sits within a family of related, mutually-reinforcing literacies rather than as a standalone skill.

European AI competence frameworks

The proposal is anchored in existing European frameworks:

• The EU Digital Education Action Plan (2021-2027) identifies digital skills for "literacy, including combating misinformation, teaching computer literacy, [and] knowledge of data-intensive technologies."
• de la Higuera's five pillars for building an AI curriculum: (1) Data Awareness, (2) Uncertainty and Randomness, (3) Coding and Computational Thinking, (4) Critical Thinking, and (5) Post-AI Humanism — the idea that AI forces reconsideration of fundamental truths about the human being (Truth, Experience, Creativity, Intelligence).
• DigComp 2.2 (Italian Digital Agency, 2022), which addresses new and emerging AI technologies including personal data, interaction with AI systems, IoT, environmental Sustainability, new forms of work, virtual/augmented reality, and robotisation.

Two levels of AI in education

The paper distinguishes two ways AI enters teaching:

• Direct didactics of AI — fostering knowledge of AI definitions, vocabulary, fields of application, and modes of operation; essentially building AI literacy so students use these technologies consciously (e.g., learning how machine learning works, how AI is trained, how data is used).
• Indirect teaching of AI — leading students to recognize and use AI applications to solve a problem or achieve a goal, learning to collaborate with AI to improve the teaching/learning process.

Both levels operate under the educational mode of Explainability, in which a teaching activity is designed within a soliciting educational environment centered on social interactions, initiating a process of "AI culture" in the school community.

The Episode of Situated Learning (ESL) instrument

ESL is an active teaching instrument built on three verbs — anticipate (students elaborate initial pre-knowledge and form hypotheses), produce, and reflect — that functions at the micro level (lesson planning) and macro level (curriculum construction). The authors propose declining these two levels for developing an AI Education curriculum. ESL is circumscribed temporally and in content (its "episode" nature, consistent with microlearning logic), begins with a preparatory phase enabling cognitive anticipation, and supports autonomous questioning and hypothesis-formation.

Educational significance

This paper connects AI literacy to situated learning theory as a curricular design strategy. It positions AI competence as something to be developed through active, situated episodes rather than abstract instruction, bridging technical and non-technical disciplines. The "Post-AI Humanism" pillar links to the broader philosophy of AI in education discussion, while the direct/indirect distinction and explainability mode connect to Learning Design and K-12.

Key Findings

• Proposes a curricular AI literacy framework built on the Episode of Situated Learning (ESL) instrument (anticipate, produce, reflect), usable at micro (lesson) and macro (curriculum) levels.
• Situates AI literacy within the multiliteracy landscape: media, information, data, and AI literacy.
• Grounds the framework in European AI competence frameworks (EU Digital Education Action Plan, de la Higuera's five pillars, DigComp 2.2).
• Distinguishes direct didactics of AI (learning how AI works) from indirect teaching of AI (learning to collaborate with AI), both under an "explainability" mode.
• Emphasizes the Post-AI Humanism pillar — AI prompting reconsideration of human truths (truth, experience, creativity, intelligence).
• Argues for interdisciplinary bridging between technical and non-technical disciplines.

What this means for practice

• Instructors. Plan AI Literacy as bounded, timetabled episodes that move through anticipate, produce, and reflect, using the opening phase to surface students' existing pre-knowledge and hypotheses about AI before any tool is used.
• Instructors. Label which of the two jobs an activity serves — direct didactics of AI (definitions, how a model trains on data, its modes of operation) or indirect teaching of AI (using AI to solve a problem and collaborating with it) — and make sure students meet both.
• Instructors. Anchor episodes in real-world problem-solving and social interaction and run them in an "explainability" mode, so students interrogate how a tool works and where its limits lie rather than only consuming its output.
• Instructors. Reserve reflection time for the post-AI humanism questions — what these systems force us to reconsider about truth, experience, Creativity, and intelligence — since the framework treats that pillar as part of the competence, not an add-on.
• Administrators. Build the AI curriculum at the macro level as a sequence of episodes that bridges technical and non-technical disciplines, rather than as a standalone technology module bolted onto one subject.

Limitations

• This is a conceptual proposal in a CEUR workshop paper: the ESL-based AI framework has not been field-tested, and the authors state that its validation "requires an extended course of field experimentation."
• The competence structure is derived from European policy sources — the EU Digital Education Action Plan (2021–2027), DigComp 2.2, and de la Higuera's five pillars — rather than from classroom observation, so the framework inherits their scope and priorities.
• ESL was developed as a general active-teaching instrument for other content; the paper offers no evidence about how well anticipate–produce–reflect maps onto AI-specific objectives or about how episodes should be sequenced across grades at the macro level.
• The claim that both levels of AI teaching operate under an "explainability" mode centered on social interaction and school-community participation is asserted without criteria for judging whether any given episode achieves it.

Citation

Panciroli, C., Allegra, M., Gentile, M., & Rivoltella, P. C. (2023). Towards AI literacy: A proposal of a framework based on the Episodes of Situated Learning . CEUR Workshop Proceedings (Ital-IA 2023).