Method for evaluating artificial intelligence tools

Explanatory note

1. Purpose of the assessment

This grid aims to evaluate AI tools that can be used in a general or educational context in Switzerland in a neutral, systematic and contextualised manner, taking into account national requirements in terms of data protection, security, ethics, educational relevance and sustainability.

2. Rating system and Swiss context

Each criterion is rated on a scale of 1 to 5, according to a standardised scale:

1 = Very low / No evidence or implementation
2 = Low / Generic mention or partial implementation
3 = Average / Basic compliance, functional but limited
4 = Good / Well implemented, reliable, with documentation or evidence
5 = Excellent / Exemplary, with external audit, certifications or proven institutional adoption

This scale allows for an objective comparison between tools, while reflecting the high standards of the Swiss context, which is based on three fundamental principles:

Strict compliance with the nLPD, in particular for sensitive data;
Preference for solutions hosted in Switzerland or the EU, with rejection of transfers to high-risk countries;
Alignment with the values of the Swiss education system: inclusion, accessibility, neutrality, transparency and digital sobriety.

3. Evaluation criteria and objectives

Axis 1: Credibility and reliability of the supplier

Axis 2: Data, security & compliance

Axis 3: Reliability, explainability & bias

Axis 4: Ease of use & relevance for teaching

Axis 5: Environmental and ethical issues

4. Scope and limitations

The assessment covers the current version of the tool (free and/or professional);
The assessments are neutral and factual;
They may evolve with tool updates, regulatory changes (e.g. AI Act) or constructive feedback from the HES-SO community via the dedicated form.