Does the Outline Agreement Take a New Digital Path?

New Ambitions for Repackaged Old Policy

The outline agreement addresses AI and digitalisation at 16 points. We have tried to rank these according to the extent to which they continue existing policies versus responding to recent developments and new challenges.

Latest Ambitions

• Scientific standards for models and algorithms: Developing open and imitable standards with clear package inserts is a relatively new ambition that responds to recent concerns about the transparency and reliability of AI.
• Resilience against disinformation and deepfakes: The explicit emphasis on protecting society against disinformation and deepfakes is a new development arising from recent technological and societal challenges.
• Specific focus on economic espionage and electronic equipment from high-risk countries: This is in response to recent geopolitical tensions and technological risks.

Partly New Ambitions

• Innovation and rapid legal application: Although innovation has always been a policy goal, the emphasis on rapid legal application and flexible adaptation of procedures is a refinement of existing policies.
• A special innovation programme for robotisation: Establishing a specific programme for labour-saving robotisation is a new focus within the broader context of technological innovation.
• Increasing maximum penalties for cybercrime: Increasing penalties is an existing policy that is now being intensified in response to increasing cyber threats.
• Investigation of a security organisation such as DGSI: This is a new specific policy measure, although it fits within the broader framework of national security.

Continuation and Refinement of Existing Policy

• Digitalisation of the NPO: Although digitalisation is an ongoing trend, the focus on reaching a younger audience is an intensification of existing goals.
• Improved accessibility of government services: This builds on existing efforts to make government services more accessible, but with new emphases such as enforcement of non-digital communications.
• Use of ANPR cameras in border regions: The use of digital investigation tools is a continuation of existing policy, with a specific focus on border regions.
• Faster justice through digitalisation: This is an ongoing policy goal that is continued and intensified by technological advances.
• Stricter measures against football hooligans: Although measures against hooligans already exist, the introduction of a digital reporting obligation and a ban on facial coverings are reinforcements of the existing policy.
• Strengthening the digital capabilities of intelligence services and cybersecurity in defence: This is a continuation of existing policies that are continuously adapted to new digital threats.

Repackaging Old Policy

• Balance between innovation and security/privacy: Although stated more explicitly, the balance between innovation and security/privacy is an ongoing challenge and not a new ambition.
• Collaboration between government, science, and business: This is an existing policy line that is now receiving extra emphasis in the context of digital threats and cyber security.
• Updating the WIV: Adjusting the Intelligence and Security Services Act (WIV) is an ongoing policy that is now being specifically updated.

Colouring Existing Vision on Generative AI

A comparison of the main outline agreement with the current government-wide vision on generative AI shows that the agreement focuses more on new challenges and refines existing policies to better fit the current technological and social context.

The agreement includes new focus areas that respond to recent technological and societal concerns, such as the transparency of AI models, protection against disinformation, and the risks of economic espionage. The agreement places an explicit emphasis on new technological and geopolitical challenges, with specific measures to address them.

Existing objectives are further refined and expanded with specific programmes and measures, such as increasing penalties for cybercrime and establishing new security organisations.

Finally, the agreement places a specific focus on new accents within existing objectives, such as the digitalisation of government services and the use of technology in investigation. With this tightening, the agreement responds better to current needs and technologies.

Strong Agreement with EU Initiatives

The three latest ambitions from the outline agreement show strong similarities with broader EU initiatives. The Netherlands aims to develop open and imitable standards for AI models and algorithms, with an emphasis on transparency and clear documentation. This ambition is closely aligned with the EU AI Law, which promotes transparency, accountability, and security in AI systems. Both focus on creating trustworthy AI.

The main agreement focuses on protecting society against disinformation and deepfakes. This is in line with EU efforts, such as the Horizon 2020 projects, which develop tools and frameworks for detecting and countering disinformation. The EU approach includes promoting media literacy and improving origin tracing of digital content.

The Netherlands has a specific focus on economic espionage and the investigation of electronic equipment from high-risk countries. This ambition reflects the EU's concerns about economic security, highlighted by the EU's response to global risks such as the COVID-19 pandemic and Russia's invasion of Ukraine. Both recognise the need for greater resilience in supply chains and reduced dependence on high-risk countries.

However, the Netherlands can take specific and possibly stricter measures to better address local challenges. This can lead to faster implementation of certain technologies and policies within the Netherlands compared to the broader, sometimes slower-moving EU frameworks. Where the coalition wants to go beyond EU initiatives is not stated in the main lines agreement.

Challenging Agenda

Most of the ambitions of the outline agreement require significant efforts in technical development, cooperation, financing and policy making. The success of these ambitions depends on the availability of sufficient resources, expertise and the willingness of various parties to work together on these complex challenges. The further elaboration in policy by the new cabinet will determine the realistic implementation of these ambitions. The main challenges are:

• Balance between security and privacy: Finding a balance between enhanced security measures and safeguarding privacy rights remains a complex challenge. This includes promoting innovation without endangering citizens' privacy, ensuring due diligence in legal processes, expanding security services without violating civil rights, and maintaining freedom of information in combating cyber threats.
• Technical and legal complexity: Developing and implementing AI standards and regulations requires significant expertise and interdisciplinary collaboration. This requires in-depth knowledge of both the technical aspects of AI and the legal frameworks needed to effectively regulate these technologies.
• Rapid technological advancements: Keeping up with the rapid evolution of technologies such as deepfakes and AI algorithms is an ongoing challenge. Policymakers must continuously innovate and adapt to effectively manage new risks and opportunities.
• Resources and financing: Sufficient investments and resources are crucial for realising AI ambitions. This includes financial support for research, development and implementation of AI technologies.
• Collaboration and coordination: Effective collaboration between different national and international parties is essential for success. This requires coordination between governments, industry, academic institution, and international organisations to achieve common goals.
• Publicity versus security: The pursuit of transparency can conflict with security needs. Detailed disclosure of models and algorithms can facilitate abuse by malicious parties, necessitating a balance between disclosure and security.
• Data centralisation versus decentralisation: Centralisation of data can conflict with the need for decentralisation to improve privacy and data security. It is crucial to balance the efficiency and control of centralised data storage with the security and privacy of decentralised systems.

Ambitions That Do Make a Difference

Generalities do not break new ground, while more precise objectives do. The outline agreement shows that this is possible for topics such as deductibles in healthcare and asylum. The current government vision for AI is characterised by broad and general objectives. The agreement could have broken new ground by formulating the objectives in a more concrete and ambitious manner:

• Scientific standards for models and algorithms: All AI systems developed or implemented within the Netherlands meet the requirements of the EU AI law, with a compliance rate of 100%, audited by annual audits and reports in collaboration with relevant stakeholders.
• Resilience against disinformation and deepfakes: By the end of 2025, the impact of disinformation campaigns will be reduced by 50% through a national protection programme.
• Specific focus on economic espionage and electronic equipment from high-risk countries: By the end of 2024, all government agencies and major companies will conduct risk assessments and implement security protocols.
• Innovation and rapid legal application: By the end of 2025, the average lead time of legal processes will be reduced by 30% due to digital tools and flexible legislation.
• Special innovation programme for robotisation: Ten pilot projects for labour-saving robotisation were started at the end of 2024, aimed at a 30% productivity improvement.
• Improved accessibility of government services: Citizens' satisfaction scores with government services will have improved by 25% by mid-2025 due to improved accessibility.
• Faster justice through digitalisation: By the end of 2025, the average processing time for legal cases will have been shortened by 30% due to further digitalisation.
• Digital inclusion: By the end of 2026, 95% of the Dutch population will have access to digital services, regardless of their technological skills or socio-economic status, by offering free digital skills training and providing subsidised internet access for low income groups.

Missed Opportunity

Although the outline agreement requires significant investments, it also includes significant cuts and a reduction in the number of government civil servants. Here lies an opportunity for AI to play a key role in increasing productivity, shortening lead times, reducing costs, and improving government services.

Why the Government Should Prioritise AI »

The appendix states that the coalition aims for an accelerated upscaling of digital care and thus increases labour productivity. This is also possible in other areas under the right preconditions. The current provisional position for government organisations when using generative AI emphasises, however, caution and compliance with laws and regulations. Innovation is stimulated through experiments and pilots within secure frameworks, while non-contracted AI such as ChatGPT is usually not allowed. The outline agreement could have been more ambitious by aiming for implementation of AI in at least five government services by the end of 2025, with a 20% efficiency and quality improvement, measured through KPIs such as time savings and customer satisfaction.

In addition, the coalition could have protected AI projects with a minimum return on investment (ROI) of 200% from cuts. There is also a small cut, €13.5 million, that can be achieved by eliminating the Dutch government's development of its own generative AI model. Becoming less dependent on American tech giants such as OpenAI will have to be tackled, at least at the EU level, if you have any chance of success. The cheaper and easier way is to stimulate more competition.

PVV, VVD, NSC, and BBB could also have drawn inspiration from countries such as Finland and Estonia by placing AI more centrally in the chapter on good governance. With this, the Netherlands would have truly broken new ground.

Additional Ambitions

The main agreement is not complete in the field of AI and digitalisation. In addition, existing policies could have been made more concrete and measurable. Some examples:

• Education and training in AI
  • AI in education: By the end of 2025, AI education must be integrated into the curriculum of at least 80% of primary schools, secondary schools, and universities in the Netherlands. This is measured by the number of educational institutions offering AI courses and the number of students taking these courses.
  • Upskilling the workforce: By the end of 2025, AI upskilling programmes should be available to at least 50,000 workers to improve their digital skills. This is measured by the number of training sessions completed and the skills assessments of the participants.
• Ethics and responsibility in AI
  • Ethical Framework for AI: A national ethical framework for AI development and use should be developed and published by the end of 2024. Success is measured by the adoption and compliance with the framework by at least 80% of AI companies and institutions in the Netherlands.
  • AI Ethics Committee: An AI Ethics Committee should be established by mid-2025 and publish at least two reports annually on compliance and recommendations. This is measured by the number of reports published and the implementation of the recommendations.
• Partnerships for innovation: It will not be easy to formulate a credible ambition now that major cuts are being made in multi-year programmes, such as the phasing out of the National Growth Fund. Yet the government will not be able to do it alone.
• Social applications AI: By the end of 2027, at least ten AI applications should be developed that address societal challenges such as healthcare, environment, and education. This is measured by the number of subsidised projects and the impact measurement of these applications.
• Diversity programmes: By the end of 2025, programmes should be developed and implemented to increase the participation of underrepresented groups in the AI sector by 20%. This is measured by the number of participants from underrepresented groups and their progression through the sector.

Conclusion

The Outline Agreement 2024 - 2028 also focuses on digital innovation and the integration of AI in various policy domains. Our analysis shows that the agreement mainly builds on existing policies and responds to current technological and societal challenges such as disinformation and economic espionage. However, the policy agenda for AI and digitalisation is presented in a fairly general and fragmented manner in the agreement. Success will depend greatly on ambition, feasibility, and available resources. The actual impact of this agreement remains uncertain, as it depends on effective implementation and cooperation, both within the EU and internationally. It can therefore be said that this agreement shows little courage in the field of AI and digitalisation and unfortunately does not take an innovative path.

Turn Your AI Ambitions Into Reality

With our consultation, you can realise the full potential of AI. The outline agreement establishes the foundation for AI and digitalisation, but meaningful innovation requires strategic direction and flawless implementation. Why Symbio6? We create AI plans that are tailored to your objectives and keep up with cutting-edge AI technologies, while ensuring that all AI projects fulfil regulatory criteria. Our package of services is intended to guide organisations through every stage of integrating AI into their operations, whether they are trying this technology for the first time or looking to upgrade their existing AI systems. Collaborate with us to turn your AI vision into reality. Contact us today to schedule a consultation.