Reflections from a European conference on data standards

The SEMIC 2023 conference—focused on ”Interoperable Europe in the age of AI”—brought together a few hundred politicians, policymakers, analysts, and even ontologists interested in data standards and interoperability last month in Madrid, Spain. The conference showcased the importance of public sector leaders valuing the data and is closely related to the work promoted by the Beeck Center’s networks. Below are my main takeaways.


Data governance is a leader’s job

Siim Sikkut, the former chief information officer of Estonia and managing partner of digital change advisory Digital Nation,  centered his keynote address on data governance as a leader’s job. Leaders must understand how to give meaning to data, produce metadata, and care about data quality in order for their countries or organizations to advance their missions. While they do not need to become technical writers or experts on what each table or column means, they do need to provide resources and prioritize such work. No technical solution will work unless the data is valid, accurate, understood, accessible, and interoperable.

Sikkut’s vision for Estonia’s digital future is to have an app, Burokratt, that simplifies people’s lives by proactively sending them reminders to renew their passports and allowing them to accomplish renewals on their phones, among other tasks. It is a platform of automated workflows, data integration, and a user interface. 

Sikkut reminded the audience that digital projects like these ultimately require good data, strong data standards, and interoperability—the ability of software and systems to exchange data. He also urged leaders to resource action plans well, put teams in place, and dedicate actual work to achieve ambitious digital plans. 

Eileen Fusch, a representative from Germany’s Ministry of the Interior and for the Community, described Germany’s creation of a national data catalog during an event panel. She stressed that data engineers or other technologists cannot provide meaning to the data alone; they need to talk to people who interact with data and provide actual public services. Dedicating space and time to finding the right people to talk to in order to capture the accurate meaning of data is essential to a data project’s success.


AI is an assistant, not a deputy

While many countries are cautious of using artificial intelligence (AI), Romania is experimenting in using an AI-powered structure, Ion, to gather input directly from residents and about government services.

The government sees value in this virtual advisor, but they treat it as an assistant, not as a deputy. At this time, there is a prototype enabled with natural language processing (in Romanian) and it is being taught to interact meaningfully with citizens. While Ion is somewhat controversial, the government is planning to place it around the country to take questions and suggestions from residents, process the input, and provide analysis to public administrators. As Eduard Mititelu, a representative from Romania’s Ministry of Research, Innovation and Digitalization, said, “one person is not able to have so many conversations at the same time across the country.” 

Panelists also argued for radical simplification of public services with the help of technology and machine learning. For example, Spain used to request about 60 documents when an entrepreneur requested a subsidy. Now, with the use of technology and AI, the government does not request any documents because the information and data points can be pulled from other databases. Similarly, in Ukraine, ministries are competing to decide which service will be added to the popular Diia app, which allows citizens to access digital documents.  A representative from Ukraine’s Ministry of Justice shared that seven out of 10 entrepreneurs registered their businesses via the app, saving what would have amounted to 72 years of people’s time if traditional processing of the paper registration forms had been used.

Finally, panelists also mentioned the importance of the use of voice and interpersonal interactions between citizens and government. Talking is a natural way for people to communicate, and several European countries are experimenting with allowing people to request public services by just speaking (instead of filling out forms) in their own language of preference. 


The times have changed

The mood at the conference stressed that the times have changed and public administration must change as well. With deployment of large language models and different expectations from citizens, public administrators cannot ignore the changes; instead they must become more articulate in the data management details and hire more ontologists in order to scale up the digital service pilots and master the changing landscape. An example from Denmark summarizes the situation well. In 2020, the country’s government noticed that they had not had any physical bank robberies. However, this was not a celebration moment for Danes —all robberies had moved online and now they have to outsmart and catch cyber criminals. 


Additional European Union data resources:

  • European Union data strategy
  • European Union Data Governance Act
  • European Union registry of semantic models
  • Proposed European Union Artificial Intelligence Act