Built-in AI*
The artificial intelligence built into Cogitare is not a tool for speeding up students' work — it is a guide designed to help them think further, never to think for them. Where most everyday uses of generative AI involve asking a question and getting an answer, Cogitare's assistant does the opposite: it asks questions, using a Socratic approach, to push students to surpass themselves — never ready-made answers.
What It Sees, and What It Doesn't
The assistant only has access to the context strictly necessary for its educational role — never to a student's entire file. In the Projects module, only the pitch of the current project is shared with it. In Academia, its view widens slightly to take the progression of learning into account, with a way of questioning suited to that different context.
In both cases, the targeted competencies and the feedback already given by the teacher inform its questions, so that they extend human guidance rather than ignore it.
Safeguards
Artificial intelligence in a school setting calls for particular care. Cogitare applies clear principles:
- No validation without questioning. The assistant does not flatter, does not validate blindly, and never simulates an emotional bond with the student.
- Limits on use. The context shared remains strictly educational, with no broader, irrelevant access.
- Human vigilance, never automation alone. If a sign of distress is detected in an exchange, the student's tutor teacher is informed — it is never handled by the AI alone.
- Constant transparency. Students always know they are interacting with a program, not a person.
- Human oversight. Exchanges remain available to the tutor teacher, and the assistant's behaviour is reviewed periodically.
Why this approach
Poorly governed AI in education carries a well-documented risk: doing the cognitive work in the learner's place, undermining the very competencies it claims to support. Cogitare was designed to avoid exactly this trap — here, AI serves the student's autonomy and never replaces it.