A confidence that is borne out

A citizen-driven tool must be open to review, criticism, and correction. Here are the risks to take seriously and the safeguards that HUMANIA aims to establish.

Accessible sources, correctable errors

AI can produce inaccurate answers or omit essential context. Analyses must refer to the documents used, specify their date, and indicate any uncertainties. Citizens must be able to request corrections.

Multiple points of view

Data and methods can reproduce biases. The goal is to examine arguments according to explicit criteria, to present disagreements, and to distinguish between facts, interpretations, and opinions.

A human responsibility

An automated recommendation is not a democratic decision. Those who use or publish an analysis remain responsible for their choices. Elected institutions retain their role.

Transparency with respect for privacy

Making methods public does not mean exposing personal data. The project must limit data collection, explain its uses, and protect contributions that are not intended for public use.

Honest consultations about their limitations

Each survey must specify the question, the period, the number of responses, and the methodology. Voluntary participation alone cannot claim to represent the entire population.

An open and progressive control

Documentation, independent assessments, and citizen participation in governance are among the objectives. Their implementation must be clearly described: any announced guarantee must be verifiable in practice.