Responsible Use of Artificial Intelligence in Scolaro

Scolaro uses artificial intelligence (AI) to reduce teacher workload, not to replace them.

This page explains, in simple language:

  • how AI is used within the platform;
  • what AI does and does not do;
  • how we process student and teacher data;
  • what are the known limits and risks of these technologies;
  • why a framework like Scolaro is more secure than using AI tools "secretly."

1. How AI is used in Scolaro

In Scolaro, AI acts as a pedagogical assistant. It never makes decisions about evaluations or a student's progress on its own.

1.1 Generating tests and questions

  • The teacher chooses the level (Sec 1 to 5), the type of text or task, the targeted skills and, if necessary, imports documents (course notes, PDF, texts, etc.).
  • Based on these elements, the AI proposes a draft evaluation: development questions, multiple-choice questions (MCQs), writing tasks, multiple versions, etc.
  • The teacher can then modify, reorganize, or delete any question before publishing the test.

AI never creates an exam "by itself" without human instruction or validation.

1.2 Improving and creating test variants

  • AI can reformulate questions to simplify them, make them more complex, or adapt them to a group.
  • It can generate A/B versions of the same exam (same requirements, but different order or numbers).
  • The teacher maintains control over the final version of the test at all times.

1.3 Rubrics and marking schemes (C1, C2, etc.)

  • Scolaro uses AI to propose criterion-referenced evaluation rubrics (e.g., C1 – understanding, C2 – organization, etc.), inspired by practices in French language instruction at the secondary level.
  • Descriptors, weighting, and mastery levels are generated based on the grade level and the text type.
  • Teachers can adapt the rubric to their reality: modify criteria, change points, disable elements.

1.4 Assisted grading

For certain development questions or writing tasks:

  • the AI proposes a suggested grade, with a marking scheme and a brief justification;
  • the teacher sees the applied criteria (for example, C1–C5) and can adjust the grade as needed;
  • the teacher can ignore the AI suggestion and grade entirely by hand.

The grade recorded in Scolaro is always the grade validated by the teacher.

1.5 Text analysis and mini-corrector

  • AI can identify frequent errors (spelling, grammar, syntax, punctuation, vocabulary) in student texts.
  • It can propose a corrected version or suggestions for rewriting.
  • These tools are used to support language instruction, not to automatically penalize students.

2. The Teacher's Role and AI Limits

2.1 The teacher remains the main decision-maker

In Scolaro:

  • the teacher chooses what the AI will do (generate a test, propose a rubric, suggest a grade, etc.);
  • the teacher can accept, correct, or reject any AI suggestion;
  • the teacher is the only person who validates official evaluations.

AI is a tool, not a pedagogical authority.

2.2 What AI does NOT do

  • AI never decides alone on a student's success or failure.
  • AI does not modify official report cards or school management systems.
  • AI does not contact parents and does not take communication initiatives.
  • AI does not perform behavioral profiling (it does not process data on students' behavior, health, or private life).

3. Data Protection and Confidentiality (Pedagogical Level)

A complete privacy policy and terms of use are available here: Privacy Policy and Terms of Use.

The goal of this section is to explain, in simple terms, what happens when a teacher or student uses Scolaro.

3.1 Types of data processed by AI

Depending on the features used, the following types of data may be sent to the AI engine:

  • exam instructions, test items, and rubrics;
  • student texts (essays, long replies, etc.);
  • teaching documents (text excerpts, course notes, PDFs).

We strive to strictly limit the transmission of personally identifiable information.

3.2 Minimization and anonymization

  • We encourage teachers to avoid including personal data in text transmitted to the AI (full name, student ID, etc.).
  • When possible, Scolaro replaces direct identifiers with internal codes (e.g., "Student 1", "Student 2").
  • Content may be truncated or summarized to limit the amount of information sent.

3.3 Storage and Access

  • Tests, rubrics, and results are stored in the Scolaro database, with strict access controls (only authorized accounts have access).
  • Technical teams can access technical logs to debug and improve service quality, without using these logs to directly evaluate students.
  • Schools can request the deletion or anonymization of certain data, according to applicable legal obligations.

3.4 Transparency for schools and students

  • School administration and IT managers are informed of the features that use AI.
  • Teachers can explain to students that certain Scolaro tools use AI, notably for question generation and assisted grading.
  • Scolaro can provide, upon request, a description of the data flows associated with AI functions.

For legal aspects and obligations related to personal information (Law 25, etc.), consult our dedicated page: Law 25 Compliance.

Educational establishments should also consult:

  • the Commission d'accès à l'information du Québec (CAI): CAI Website – Law 25;
  • their own internal policies for governing personal information.

4. Bias, Fairness, and AI Limits

4.1 Known limits of AI models

  • AI can make mistakes or lack nuance, particularly in the interpretation of original responses or complex literary texts.
  • Models are trained on vast amounts of text and can reproduce biases (linguistic, cultural, social).
  • AI does not know every group, every student, or every local context; it does not replace the teacher's deep understanding of the classroom.

4.2 Measures taken in Scolaro

  • Instructions sent to the models emphasize supportive and graduated marking schemes (partial points, several acceptable responses when relevant).
  • Questions and rubrics are generated based on academic skills (not student stereotypes).
  • The teacher can correct any suggestion judged unfair, inappropriate, or biased.

4.3 Teacher's role for fairness

  • Verify that questions and texts are appropriate for their group (socio-linguistic profile, special needs, accommodations, etc.).
  • Adapt weighting or reformulate certain questions to reflect the realities of the classroom.
  • Never make high-stakes decisions (final failure, orientation, etc.) solely based on a grade proposed by AI.

5. Why Scolaro is More Secure than "Secret" AI Use

In reality, many teachers already use general AI tools (like public chatbots) to generate questions or correct texts. Often:

  • there is no clear information on what is permitted to be sent or not;
  • there is no trace of what was submitted to the external service;
  • there is no pedagogical control over how the AI evaluates or corrects.

Scolaro offers a more secure and transparent framework:

  1. Clearly defined features
    AI is used for a limited number of tasks: test generation, rubrics, assisted grading, mini-corrector. These functions are explicitly indicated in the interface.
  2. Alignment with school practices
    Proposed rubrics and marking schemes are inspired by evaluation practices used in secondary schools in Quebec. The teacher remains responsible for adaptation to their school's context.
  3. Traceability and control
    Tests, grades, and corrections are kept in Scolaro. The teacher and school can trace what was generated or suggested by AI, which is impossible with unregulated use of public tools.
  4. Constant human-in-the-loop
    No major decisions are made automatically. The AI proposes, the teacher decides.

6. External Resources on AI in Education

For establishments that want to go further in reflection on AI and education, the following resources may be useful:

These links are provided for informational purposes. Each school service centre or establishment remain responsible for applying local laws and policies.

7. Questions and Comments

School teams, administrations, and IT managers can contact us to:

  • obtain clarifications on the use of AI in Scolaro;
  • request technical information on data flows and subcontractors;
  • report a correction suggestion or content generation judged problematic.

Scolaro Contact Form

We update this page as practices, laws, and official recommendations on AI in education evolve.