Highberg supports the municipality of Rotterdam in the responsible use of algorithms and AI systems

In the role of AI-systems Expert (product owner of the algorithm registry), Highberg has taken on the identification, classification, and registration of AI systems in collaboration with the municipality. This includes providing both solicited and unsolicited advice on the algorithm governance and the responsible use of specific AI-systems. In this capacity, advice was also provided for 5 on Fundamental Rights Impact Assessments (FRIA’s), a future legal requirement for high risk systems by the AI Act.

Algorithms are employed in processes to make these processes more efficient, better, or more effective. These AI systems can be in deployment for many years, as well as new initiatives to automate certain steps in a process. AI system can be deployed in the entire organization.

Throughout this project, significant attention has been given to structurally align with various teams both to detect existing algorithms and to identify new initiatives. Through these alignments, a large-scale inventory with more than 100 interviews revealed in which processes the municipality of Rotterdam currently uses or plans to use AI systems.

Additionally, the experts from Highberg have facilitated the municipality as much as possible by incorporating and/or publishing existing applications in both Rotterdam’s and the national algorithm registry.

In executing these tasks, adherence to scope, definitions, and agreements in algorithm governance was maintained. Solicited and unsolicited advice on governance improvement led to a refinement of the algorithm governance and alignment with national and European developments, including the AI Act, the implementation framework of the Ministry of the Interior, and ChatGPT. This includes clarification of the scope (when is it considered an AI system, and when is it not) and the risk classification questions (to make them easier and consistently answerable by process owners).

Practical insights have also been applied to further automate the management tool of the algorithm registry. Examples include aligning definitions with the national algorithm registry, proposing an automated periodic check of information on algorithms, and an automatic (API) link to the national registry.

The inventories, alongside with the implementation of an internal communication calendar, have created greater awareness within the organization regarding the responsible use of algorithms and AI. Employees now have a better understanding of their roles and knowhow to approach the AI expert for questions, support, and advice.

With these activities, Highberg has established a solid foundation for the municipality on a higher level of AI literacy, making it easier for process owners to responsibly deploy algorithms and ensuring centralized documentation while improving transparency on deployed AI systems.

Carolien van Hooijdonk (team manager research and BI at the municipality of Rotterdam): “The experts of Highberg have ensured that the municipality has maintained its leading position in the field of algorithms and AI. I foresee that Highberg can carry out such activities in other organizations, allowing these organizations to benefit from the knowledge and expertise Highberg built in practice.”

Do you want to incorporate and improve the responsible use of your 'high-risk' algorithms, AI systems, or ML applications? Read more on our capability page or feel free to contact Sabine Steenwinkel-den Daas.