Epistemology of the Prompt: The Art of Asking Questions as a New School Subject

For two centuries, school has taught us to find the answers. Today, with Generative Artificial Intelligence instantly answering every question, the real cogniti

The global school system has been designed, for over two centuries, around a precise epistemological paradigm: the scarcity of answers. The teacher, the textbook, and the library held knowledge; the student's task was to learn to extract, memorize, and replicate those answers. Today, Generative Artificial Intelligence has overturned this asymmetry. We live in the era of absolute abundance of answers, where any notion, calculation, or textual synthesis is available in a few milliseconds. In this new ecosystem, the real cognitive bottleneck is no longer finding the right answer, but knowing how to formulate the right question.

The prompt – the textual instruction provided to an algorithmic model – has ceased to be a mere string of code reserved for programmers. As forcefully emerges from the most advanced pedagogical and academic literature, interacting with AI requires a deep sociotechnical competence that blends linguistic precision, rhetorical awareness, and critical thinking.

In this essay for the Scenarios and Reflections column, we will explore the urgency of introducing the epistemology of the prompt as a mandatory school subject, from primary education through university. It is not about teaching an ephemeral technical skill destined to become obsolete with the next software update, but about educating new generations in the art of structuring thought, making constraints explicit, and guiding the construction of knowledge in an unprecedented collaboration between the human mind and synthetic systems.

1. From Engineering to Literacy: The Birth of Prompt Literacy

To correctly frame the ongoing didactic revolution, it is essential to overcome the mechanical conception of human-machine interaction. We must draw a clear distinction between two approaches:

  • Technical prompt engineering: The purely syntactic optimization of the input to obtain a specific output. It is a procedural approach, focused on the machine and its operational parameters.
  • Prompt literacy: An iterative and critical sociotechnical capability. It is the ability to formulate, adapt, evaluate, and revise questions, understanding algorithmic biases and the logics that shape AI responses.

A systematic review published in SAGE Journals highlights precisely how effective prompting is not reduced to a list of commands, but requires continuous metacognitive monitoring. In this context, a groundbreaking paper published in the ERIC database ("Beyond Prompt Engineering: Prompting (L)iteracy…") uproots the term "engineering" to speak of literacy. The author argues that prompt literacy is a practice that is simultaneously iterative and intrinsically political: the user must be aware of the linguistic hierarchies, economic logics, and distributed agency hidden behind the Artificial Intelligence interface.

Teaching prompt literacy means, in fact, equipping students with the intellectual antibodies necessary to avoid disinformation, misuse of tools, and the passive acceptance of pre-packaged algorithmic answers.

2. The Epistemology of the Prompt and Rhetorical Practice

Beyond basic digital literacy, the higher concept of epistemology of the prompt stands out. This discipline represents the philosophical and structural reflection on how our questions define the boundaries of what can be known. Every prompt is an act of epistemic "framing": by deciding which words to use, what context to provide, and what constraints to impose on the Artificial Intelligence, the user establishes who holds epistemic authority and which forms of knowledge are valued.

The International Journal of Education and Social Practice positions prompt engineering as a genuine emerging form of rhetorical practice. Writing a complex prompt for an LLM (Large Language Model) involves interpretive, persuasive, and argumentative skills entirely overlapping with those required in writing a traditional academic essay. The student must declare their assumptions, anticipate the possible misunderstandings of the interlocutor (in this case, the neural network), and set verifiable quality criteria. In an era where the machine can write the text in our place, the act of writing shifts upstream: we no longer write the text, we write the conceptual perimeter within which the text must take shape.

3. Curricular Architectures: From K-12 to University

If the ability to interrogate the machine is the key competence of the 21st century, how can we integrate it into formal school curricula? The academic community is already proposing highly structured curricular frameworks.

An essay published in AI & Education Studies ("Prompt Engineering as a 21st-Century Literacy: A K-12 Curriculum Design") outlines a complete educational model based on Backward Design and Project-Based Learning. The framework includes multidimensional assessment rubrics that do not judge the quality of the final text generated by AI, but the refinement and evolution of the prompts written by the student, simultaneously addressing ethical issues such as data privacy and algorithmic biases.

In parallel, the journal Frontiers in Education structures this competence into four conceptual pillars:

  1. Understanding of structure: Mastering basic syntax (role, context, task, format).
  2. Prompt literacy: Being able to critically evaluate the algorithmic source.
  3. Iterative method: Applying reflective trial-and-error cycles.
  4. Critical online reasoning: Integrating the result into a broader and validated research process.

This structured ("scaffolded") approach, as discussed by researchers at UMass, must start from elementary schools with fundamental concepts (specificity and iteration) to culminate in universities with advanced applications of data analysis and intellectual co-creation.

4. Metacognition in the Classroom and Pedagogical Prompting

Empirical evidence shows that structured teaching of prompting positively alters learning dynamics. An illuminating study published in Taylor & Francis analyzed the introduction of prompt literacy in a mythology course for English language learners (EFL). The results are unequivocal: when explicitly taught (through task framing and the assignment of specific roles to AI), prompt literacy functions as a powerful metacognitive practice. Students use question formulation to plan their inquiry, monitor their progress, and actively revise unsatisfactory machine responses, developing "self-regulated learning."

However, for students to learn this art, teachers must first master it. The Journal of Information Technology Education (JITE) warns that prompt engineering is still too often treated as an implicit "technical magic," rather than as an explicit educational practice.

This is where pedagogical prompting comes into play: the strategic design of prompts aligned with evidence-based instructional strategies to meet specific learning needs. Progressive frameworks, such as the one proposed for Italian secondary schools in Inclusive Teaching, outline a three-stage development for teachers: basic technical acquisition (Basic), didactic deepening (Pedagogical), and creation of inclusive designs (Inclusive). The teacher does not use AI only to prepare lessons, but models aloud in class their own prompting process, showing students how to deconstruct a complex problem into a series of effective algorithmic interrogations.

Key Operational Takeaways (Takeaways for School Leaders and Teachers)

  • Assess the Process (Prompting Journals): Teachers must shift the focus of assessment. Instead of evaluating the final essay produced (which could have been generated entirely by AI in one click), they must require the submission of a "Prompting Journal." Students must document their conversation with AI, justifying why they modified a question, how they identified a hallucination, and in what way they refined the context to arrive at the optimal response.
  • Teach Epistemic Framing Models: From middle school onward, students must be trained to use frameworks such as the "Persona Pattern" (asking AI to assume a specific role) or the "Flipped Interaction Pattern" (asking AI to ask the student questions until it has enough information to complete a task). This reverses the student's passivity, transforming them into the director of the cognitive process.
  • Institutional Training on Pedagogical Prompting: Ministries of Education must invest in systematic teacher upskilling, going beyond trivial courses on basic ChatGPT use. The goal must be the integration of prompting as a tool for metacognition and didactic inclusion within existing curricular subjects (history, literature, mathematics).

Conclusions: The Art of Knowing How to Ask

We live in the paradox of an era in which we possess digital oracles capable of simulating omniscience, but we are losing the intellectual depth to interrogate them. Artificial Intelligence, left to itself, tends to provide average, flat, and statistically predictable answers. The spark of innovation, critical thinking, and true discovery lies solely in the friction generated by a question formulated in a brilliant, unexpected, and rigorous manner.

The epistemology of the prompt reminds us that asking is not an act of weakness or ignorance, but the highest form of cognitive exploration. Teaching how to manage ambiguity, make invisible contexts explicit, and demand logical evidence from a non-human intelligence is the only way not to succumb to the automation of thought.

The challenge that the global education system must face today is encapsulated in a radical question: if the future of critical thinking, democracy, and scientific innovation depends crucially on the ability to formulate questions that guide intelligent systems to produce relevant knowledge… does it still make sense for school to continue teaching our children only to find answers, or has the time come to teach them, first and foremost, the art of constructing the right questions?

Bibliographic References and Sources

Article by the Editorial Team of La Bussola dell'IA