- Open administration
- Digital Government Congress
Digital Government Congress: the space for strategic connection between the administration and companies
The director of the AOC, Miquel Estape, and the head of Innovation, Carme Callés, presented at the Smart City Expo World Congress (SCEWC) the algorithmic transparency initiative that the AOC applies to its services. The main objective is to foster trust in digital government services that use data and artificial intelligence (AI), while ensuring ethical and responsible use. The session, held last November 7 in the Ágora of the Catalonia stand of the SCEWC, was very well received. From the AOC, we would like to especially thank the representatives of Catalan councils interested in this initiative.
With this project, the AOC seeks to promote transparency, facilitate the understanding of algorithms, encourage public participation in their development and ensure quality data for training AI models. This also includes ensuring that algorithms are free of bias and are regularly monitored.
Publication of algorithmic transparency reports
To achieve these goals, the AOC has developed algorithmic transparency reports with standardized content to provide a clear and understandable view of each AI system. These reports explain, in accessible and plain language, the functioning and effects of algorithms on people. They are a tool to comply with the transparency obligations established in the regulations, as well as allowing comparison between different AI algorithms.
The reports follow international standards, such as the Amsterdam and Helsinki Algorithm Registration Model, the Eurocities European Algorithmic Standard, and recommendations from the OECD and the US NIST. These benchmarks have helped to establish good practices that facilitate meaningful transparency, allowing citizens to better understand the use of their data and the purpose of these algorithms.
Risk management methodology focused on the protection of fundamental rights
Another essential element of the initiative has been the development of its own risk management methodology to assess the impact of AI on fundamental rights. This methodology identifies fundamental principles and rights that may be affected by the use of large volumes of data processed with algorithms. The AOC thus carries out an impact assessment that allows an accessible and comprehensible risk analysis for non-experts.
This tool, agile and based on current regulations, helps to identify and mitigate risks of discrimination, data vulnerability or bias that could affect citizens.
The methodology also adapts to the requirements of the Artificial Intelligence Regulation (AI Act), which specifies the need to analyze known and foreseeable risks in relation to health, safety and fundamental rights. With this, we want to ensure that our AI systems meet the standards of security, transparency and accountability necessary to build trust in digital administration.
Here is the presentation:
To date, three reports have been published:
- AOC conversational chatbots (26 / 5 / 2023)
- Video identification to obtain the Mobile idCat (2 / 2 / 2024)
- Automation of the energy poverty report (5 / 2 / 2024)
Reports on:
- Chatbots with generative AI
- Automatic classification of incidents on public roads