Sector
Public authority
Problem to solve
The client was developing several artificial intelligence projects and adopting various tools without an explicit ethical reference point, without shared guidance and without operational indications on how to integrate (or not integrate) AI into its processes, whether the more sensitive and high-risk ones or ordinary ones. It lacked a compass for deciding which use cases to prioritize and how to oversee the most sensitive decision-making processes, as also required by the AgID Guidelines. The risk was adoption driven solely by technical or opportunistic criteria, not aligned with the organization’s values.
Goal
Define, through a participatory process, the client’s AI Ethical Vision – guiding values, ethical principles, objectives and expected medium-term impacts in adopting AI – and formalize it in a validated Ethical Code. The Code integrates the organization’s values with international AI Ethics principles (environmental and social sustainability, inclusion, non-discrimination, transparency, accountability) and with the governance mechanisms required by the AgID Guidelines and the AI Act.
Why it matters
- Defines a clear perimeter within which to use automated and artificial tools, with effective principles and practices
- Provides a shared criterion for prioritizing use cases and governing the most sensitive decisions
- Responds to the AgID Guidelines and prepares the governance required by the AI Act
- Stakeholder involvement creates ownership: the principles are not imposed from above
- Translates abstract values into an operational document, usable in everyday choices
Methodologies and frameworks adopted
- Participatory process: up to 8 semi-structured interviews (30 min) with selected stakeholders
- Analysis of internal documentation
- Alignment workshop (1 hour, max 10 participants) on orientations, needs and expectations
- Formalization and validation of the Ethical Code (document + summary sheet)
- References: international AI Ethics principles, AgID Guidelines, AI Act
Takeaways
- An ethical code works if it stems from listening, not from a template;
- Values become useful when they guide concrete choices, starting with the prioritization of use cases;
- Combining internal values, international principles and regulatory requirements in a single document avoids fragmentation;
- A lean process (interviews + one workshop) is enough to build a solid shared foundation.