Session: Practical tools and methods to implement the AI Act (EN)

Speakers: Cédric Gilissen (Wonka AI), Angeliki Tiligadi (Qover), and Jens Meijen (Umaniq).
Moderator: Wannes Ooms

Content
As the AI Act is gradually being implemented, questions arise about the practical implementation. Are companies equipped with the right tools, or are their resources falling short? A panel of AI experts and AI developers considered these questions and the existing tools of the KCDS.

Takeaways:

• Tools that translate from law to practice have a clear added value. An example is the policy prototyping project of the KCDS, which provides examples and templates, while also explaining well the logic behind the prototypes.

• Most companies are aware of the AI Act and its requirements, but struggle to find a starting point. Our panel recommends starting with AI literacy in the organisation and mapping high-risk systems under the AI Act. A good documentation culture helps with this and it is recommended to align different tools (such as the processing register under the GDPR and the register of high-risk AI systems).

• Companies can align their efforts with the timeline in which the AI Act obligations become applicable. This includes, for example, prioritising AI literacy in the first phase.

Session: Data and AI in the workplace: A win-win for all stakeholders? (NL)

Speakers: Anna-Carolina Zuiderduin (TU Delft) and Willemien Laenens (KCDS).
Moderator: Pieter Duysburgh

Content
AI and data-driven applications offer tremendous opportunities for workplaces. Yet, the actual innovation process often encounters significant challenges, with many initiatives ultimately failing. What causes these setbacks? What role does employee engagement play in this dynamic, and how can we improve the success rate of AI-driven innovation?

Takeaways:

• The future of work can evolve in many ways and depends on things like the extent to which the government takes on a steering role, how central we place labour in our lives, whether individual or collective well-being is most important, and whether technology gets a main or secondary role in this story. Scenario-based foresight can help policymakers prepare for the multiple futures of work.

• With AI innovation in the workplace, the importance of involving employees in the story is often underestimated in order to connect the technology to employees' working practices and create support for the innovation. The Data & Society Knowledge Centre developed a set of methodologies to put employee involvement into practice.

Panel: How should we organise supervision on AI in Belgium? (EN)

Speakers: Prof. Dr. Peggy Valcke (Belgian Institute for Postal Services and Telecommunications (BIPT - IBPT), Simon Verschaeren (Department of Work, Economy, Science, Innovation and Social Economy - Flemish government), and Jessica Godoy Chaparro (Digital Economy department of the Belgian Federal Public Service Economy)
Moderator: Thomas Gils

Content
This session started with the recent KCDS policy brief. We invited stakeholders to engage in a discussion with us and explain their views on how such supervision should be organised.

Takeaways:

• The debate revealed that no consensus had yet been reached on which authority would ultimately be assigned which competencies. The emphasis appears to be on the federal level, with the regions having provided little to no input thus far, despite conducting their own analyses simultaneously. Although all speakers underlined the importance of cooperation, in practice, there appears to be little communication between the different levels. The department WEWIS addressed this lack of dialogue by explicitly extending a hand to the FPS Economy, urging them to engage with the regions more quickly rather than waiting for the results of a study that is being conducted solely from a federal perspective.
• At the same time, there was broad agreement on avoiding regional fragmentation, particularly in the interest of stakeholders such as businesses, which would not benefit from an even more complex regulatory landscape. Another widely recognised challenge of regional fragmentation is the scarcity of required expertise, as the same specialised profiles would be sought by an increasing number of authorities, exacerbating recruitment difficulties.
• A notable disparity in resource availability was observed, with organisations like BIPT having greater capacity to integrate the AI Act into their work, while FPS Economy had to outsource such research due to limited resources. At the same time, each entity emphasised its own area of expertise, with BIPT asserting that it possesses sufficient in-house knowledge to address AI-related regulatory challenges, while the department WEWIS highlighted the continued importance of regional input in shaping effective oversight.
• Finally, there was no political agreement yet on the list of authorities under Article 77 of the AI Act, although preparatory work had already been undertaken. It remains clear that significant uncertainty persists regarding the final allocation of responsibilities under the AI Act.

Read the detailed report of this panel here.

Session: Flemish citizens and AI: Should everyone embrace AI? (NL)

Speakers: Professor Ralf De Wolf (UGent & KCDS), Giulia Mazzocchi (UGent & KCDS), Alenka Le Compte (Mediawijs), Neena Singh (KCDS & amai!) and Jef Van Laer (Scivil & amai!).
Moderator: Pieter Duysburgh

Content
A significant number of Flemish citizens either do not know what AI is or claim to never use it. Yet, AI systems directly or indirectly impact everyone, and AI tools can create substantial value, even on a personal level. Should every Flemish citizen be included in the AI narrative, and if so, how do we best achieve this?

Takeaways:

• The KCDS monitors show that the acceptability of AI technology is contextual, determined by the type of data, the type of actor, and the purpose. The latest monitor, on the use of generative AI in local governments, shows that local government employees especially need inspiration sessions and guidelines for use.

• Mediawijs gets the signal from the field of digital inclusion that many people of all profiles are eager to explore the possibilities of AI. There is also a lot of enthusiasm about the possibilities offered by AI for specific target groups, such as non-native speakers, the low-literate, people who speak sign language, or people who stutter. We need to guard that not only the strong benefit from the evolution in terms of AI technology.

• In recent years, the amai! program has developed a diverse and accessible range of activities, publications and tools to give a broad public insight into AI. In doing so, citizens are not only informed about AI but are also allowed to share their ideas for AI, and those ideas form the basis for an annual project call. In this way, the program ensures society-driven AI innovation.

  

  
  

Session: AI playbooks: An overview of methods to embed AI innovation in an organisation (EN)

Speakers: Emilie Maccarini (FARI), Annelies Vanderhoydonks (Digitaal Vlaanderen), Michelle Claessens (KPMG) and Annelies De Corte (KPMG).
Moderator: Jonne van Belle

Content
More and more organisations are exploring how AI can improve their processes, products and services. AI has long since ceased to be exclusively for tech enthusiasts, offering opportunities and added value across all sectors. At the same time, awareness of the risks and ethical challenges is growing, leaving many organisations balancing between innovation and responsibility. During this panel session, we discussed two methods that help organisations implement AI responsibly.

Takeaways:

• The AI Playbook (Digitaal Vlaanderen & KPMG) supports the Flemish public sector in establishing an AI strategy. It offers best practices, tools, step-by-step guidance, frameworks and checklists. Through a short survey, organisations can determine their profile and find targeted tools. The launch will be at the end of February 2025.

• ALLY (FARI & KCDS) is a guide to developing an organisation-wide governance strategy for trusted AI. With building blocks within four pillars, organisations can create a step-by-step roadmap for responsible AI use. Check out the demo via www.ally-ai.be.

• Management support is crucial to deploy AI in an active and structured way. Both methods provide concrete tools to convince higher management layers of the importance of a well-thought-out AI strategy.

Panel: Flemish AI policy plan (NL)

Speakers: Sabine Demey (Flanders AI Research Programme), Jeroen Fiers (VLAIO), Ilse De Vos (VAIA) and Rob Heyman (KCDS).
Moderator: Pieter Duysburgh

Content
The Flanders AI program has been running for five years. What have we achieved as a small region, and where do we aim to go in the next five years? This session took a broader view of the program to project ahead to 2030 and explore what will have been reached by then.

Takeaways:

• Today, AI technology is experiencing a peak. Because the Flemish AI policy plan was launched back in 2019, all players are ready to respond to the needs and opportunities today.

• The program has an ambitious spread: it targets strong AI players in Flanders, while also being there to get other Flemish companies on the AI train with it. This seems to be succeeding: recent figures show strong adoption of AI in Flanders.

• The different parts of the Flemish AI policy plan reinforce each other in several ways. The KCDS thereby cooperates with all elements of the policy plan.