The ScuolaAI Manifesto
Artificial Intelligence can extend our work — and our mistakes. We learn to ask it questions, understand its limits, check its answers, and remain responsible for our decisions. Results also depend on the model and the context: they do not measure a person’s worth. The school supports beginners and experienced learners with different forms of help and explicit criteria.
The 7 pillars
AI is a mirror, not the teacher
AI can extend our work and our mistakes. We learn to frame questions, understand concepts, and check answers. Beginners receive help defining their own goals: difficulty does not measure a person’s worth, intelligence, or morality.
No shortcuts, no needless obstacles
The Tutor alternates questions, explanations, similar examples, and hints. Support decreases as students become able to work independently. The final work remains theirs; asking for clarification, using assistive tools, or answering concisely is legitimate.
Technical skill and ethics grow together
Each level has two parallel tracks: Mind (thinking and ethics) and Hands (technical practice). The next technical level unlocks only after the corresponding ethics work is complete. We assess the quality of reasoning, sources, and consideration of consequences, including reasoned disagreement: we do not require agreement with the school’s views.
AI changes while you learn
Updates enter the curriculum through proposals, sources, and documented human review. Assessments preserve the criteria they used. Continuous automated collection and a RAG pipeline are planned developments: we do not claim real-time verification that is not operational.
Learning also happens through exchange
In group projects, learners discuss sources, disagreements, and decisions. Each member accepts the invitation, declares their contribution, and acknowledges others’ work. Groups are formed through invitations and human support; automatic matching remains to be developed. Timing and participation arrangements are agreed on when accommodations are needed.
Traceable authorship: thought has an author
Every submitted project clearly declares what you thought through yourself, what AI amplified, and what you built with the group. Required AI-free moments — just paper and thought — exercise originality. Assessment includes the person’s reasoned contribution, not only technical quality. Versions and editor traces document changes: on their own, they neither establish the origin of ideas nor detect plagiarism.
Generative reciprocity: those who learn, teach
Enrolled members can ask for help and suggest improvements. Those supporting other students need preparation and guidance from a supervisor. Council, Alumni, and public access to proposals are planned developments, not services already available. Contributing is an opportunity: it does not require sharing intimate experiences, and it is not a price you pay to receive help.
What we are not
- We are not an online course. In-person participation is required because thinking develops through exchange with others.
- Prompting is one part of the method: learning to state requests, criteria, and constraints goes hand in hand with knowledge, experimentation, and verification.
- We are not a program that turns out identical certificates. The goal is to develop a distinctive contribution, not to standardize people.
- We are not a closed institution. We are a community that grows with its members.
The outcomes we work toward
The program aims for these outcomes, to be verified through each student’s work:
- real, current technical proficiency — not repackaged courses from three years ago.
- a voice of your own: a portfolio with clear authorship and a traceable process, distinct from anyone else’s.
- a network of peers you have actually worked with, not simply exchanged LinkedIn profiles with.
- a way to stay current without starting over every time AI changes.
Transparency and the right to ask questions
You know the criteria before an assessment. Afterward, you can read the evidence and reasoning, see whether the assessment came from AI or a person, and request human review. A technical failure is not a judgment on your learning. No character or message count can establish understanding on its own.
The digital core is operational; the educational format and its outcomes are under validation. Declared independent attempts, comparisons between assessors, and delayed review help establish what learning lasts. We do not present an uncompleted pilot as an achieved result.