Responsible Use of Artificial Intelligence in Scolaro

Responsible Use of Artificial Intelligence in Scolaro

Edited in March 2026

1. Guiding principles

Scolaro uses AI to reduce teacher workload, not to replace teachers. Our approach is built on transparency, human control, and data protection.

For the legal and privacy framework, also consult our privacy policy and terms of use and the Law 25 page.

AI is a pedagogical assistant. The teacher always has the final say on content, grades, and decisions that affect students.

2. Permitted uses in Scolaro

Test generation and variants
  • The teacher selects the level, targeted competencies, task type, and reference documents.
  • AI can propose a draft assessment, reformulate questions, or generate equivalent versions.
  • The teacher can edit, reorganize, or remove any proposal before publishing.
Rubrics, marking schemes, and assisted grading
  • AI can propose criterion-based rubrics, marking schemes, and an initial suggested score.
  • The teacher can adjust, reject, or fully redo any suggestion.
  • The score stored in Scolaro is always the one validated by the teacher.
Text analysis and feedback support
  • AI can help identify frequent errors or propose rewriting suggestions.
  • These tools support learning and revision rather than automatically penalizing students.
  • Suggested comments and corrections should always be reviewed in their pedagogical context.

3. Human control and limits

What the teacher controls
  • The teacher decides when AI is used and for which task.
  • The teacher can accept, correct, or reject any suggestion.
  • The teacher remains the only person who validates official evaluations.
What AI does not do
  • AI never decides on a student's success or failure by itself.
  • AI does not modify official report cards or school systems.
  • AI does not contact parents and does not perform behavioural profiling.
Known model limitations
  • Models can make mistakes, miss nuance, or reproduce certain biases.
  • They do not know the full context of every classroom or each student's specific needs.
  • High-stakes decisions should never rely only on AI output.

4. Data and confidentiality

Types of data processed
  • Depending on the feature used, AI may process instructions, questions, rubrics, student texts, or pedagogical documents.
  • We try to strictly limit the transmission of directly identifiable personal information.
Minimization and anonymization
  • We encourage teachers not to send full names, student numbers, or other unnecessary identifiers.
  • Where possible, Scolaro replaces direct identifiers with internal codes.
  • Content may be truncated or summarized to reduce the amount of data sent.
Storage, access, and transparency
  • Tests, rubrics, and results are stored with strict access controls.
  • Technical teams may review technical logs for debugging and service quality, without using them to directly evaluate students.
  • Educational establishments can request more detail about data flows, deletion, or anonymization operations.

5. Fairness and usage framework

Measures to reduce bias
  • Instructions sent to models favor graduated marking schemes and school-appropriate phrasing.
  • Questions and rubrics are generated from academic competencies, not stereotypes about students.
  • The teacher can correct any suggestion considered unfair, inappropriate, or biased.
Why this framework is safer than unstructured AI use
  • AI usage is limited to clearly identified pedagogical functions.
  • Content and suggestions remain traceable inside a structured environment.
  • A human remains in the loop to validate what matters most.

6. Questions and resources

School teams, administrators, and IT leads can use the contact form to ask questions about AI usage in Scolaro or report a problematic suggestion.

For additional context, you can also consult the Commission d'acces a l'information du Quebec and other institutional resources on AI in education.