Proxima ZSP

It does not give answers. It trains reasoning.

Proxima ZSP is designed to support learning through maieutic dialogue. When a student asks for a solution, the system does not directly provide the final answer. Instead, it guides the student through targeted questions, reformulations and requests for explanation.

The goal is not merely to reach the correct answer, but to make the cognitive process behind that answer observable: what the student has understood, where the reasoning breaks down, which intermediate steps are missing, and whether the same reasoning can be transferred to other problems.

Why it was created

Generative artificial intelligence has already entered schools. Not always through the front door, often through a side entrance: automatically generated homework, ready-made answers, summaries, essays, translations and solved exercises.

The problem is not simply that students may “cheat”. The deeper problem is that, when used poorly, AI can shrink the space for personal reasoning.

Proxima ZSP starts from a different hypothesis: AI can be designed not as a cognitive shortcut, but as a learning environment. Not a tool that thinks instead of the student, but a tool that helps the student think better.

The Zone of Proximal Development

The name Proxima ZSP refers to Lev Vygotsky's Zone of Proximal Development (ZPD): the space between what a student can do independently and what the student can do with appropriate guidance.

In Italian, this concept is known as Zona di Sviluppo Prossimale, abbreviated as ZSP. The product name intentionally keeps the Italian acronym, while the English page uses the internationally common expression Zone of Proximal Development (ZPD).

Proxima ZSP attempts to translate this principle into an interaction with artificial intelligence. The system does not simply classify an answer as correct or incorrect: it supports the student's path inside that intermediate area where learning can actually take place.

The L×M×C framework

Proxima ZSP is connected to the L×M×C framework, developed by Marco Iannacone and published in open science as a pre-validation proposal.

The framework proposes criteria for observing and evaluating the quality of students' reasoning during interaction with maieutic artificial intelligence systems.

• L · Depth of reasoning
• M · Metacognition
• C · Consolidation over time and across domains

L observes whether the student proceeds by trial and error or builds a testable hypothesis. M observes whether the student is aware of their own process: whether they can say where they got stuck and what they already tried. C observes whether the same quality of reasoning reappears days later, even on a different problem: the difference between an isolated flash of insight and a competence that survives over time.

A project open to criticism

The L×M×C framework has been published in open science to encourage discussion, criticism and improvement by the scientific, educational and institutional communities.

The decision to publish it openly comes from a precise conviction: if artificial intelligence is going to become a stable part of educational processes, we should not allow only commercial platforms to define what “learning” means.

We need tools, criteria and frameworks that can be publicly discussed. And criticized, naturally. That is how science works: someone makes a proposal, others check whether it stands, and then everyone pretends not to be personally offended by the revisions.

Within the debate on algorithms that teach thinking

Proxima ZSP belongs to the broader debate on how artificial intelligence can promote critical thinking, active learning and cognitive autonomy.

In an article published on Formiche.net, Marco Iannacone connected Proxima ZSP to Professor Mario Caligiuri's reflection on algorithms capable of teaching people to think.

The reference to Caligiuri concerns the cultural and political framing of the topic. Proxima ZSP and the L×M×C framework were instead conceived and developed by Marco Iannacone.

What it can be used for

Proxima ZSP can be used as a basis for experimenting with new AI tutoring models in schools and educational settings. The system is not designed to replace teachers, educators or human tutors. It is designed to offer a structured environment in which AI can help make student reasoning more visible.

Who designed Proxima ZSP

Proxima ZSP was designed by Marco Iannacone, an Italian independent researcher, technologist and entrepreneur with experience in EdTech, open-source, internet infrastructure and cybersecurity.

Iannacone previously founded and developed EdiTouch, a suite of educational applications for students with special educational needs and specific learning disorders. EdiTouch was validated through experimentation in Italian schools and also recognized at European level as an inclusive educational innovation experience.

With Proxima ZSP, he proposes a different direction for the use of artificial intelligence in learning: not a machine that provides answers, but a system that helps students build thought.

ORCID profile · LinkedIn