News item | 07-10-2022 | 15:30
The government will quickly take measures to prevent citizens from getting into trouble as a result of incorrect use of algorithms. State Secretary Van Huffelen (Digitalization) writes this in a number of letters to the House of Representatives. The appointment of an algorithm supervisor, transparency about the use of algorithms and implementation of effective European regulations are spearheads in this regard.
Citizens must be able to keep control of their digital life and be able to trust the digital world. That is why it is important that algorithms that come onto the market respect people’s rights. This applies to algorithms of companies, but also to applications by the government. The government has an exemplary role in this regard.
Van Huffelen: “When using algorithms, the emphasis should not only be on punishing people. There are now also many algorithms that do good things. Especially now that people are experiencing financial difficulties, I want the government to mainly use algorithms to help people, for example when using schemes such as the UWV and SVB. That there will be an algorithm that points out schemes that they are missing out on to which they are entitled. Now I’m going to find out how something like that could work. I have also allocated money to come up with solutions to fight poverty and problematic debts with data and algorithms. Algorithms should only be used when it is really necessary for the government to work properly.”
Cashing in on opportunities
Governments have a duty to protect fundamental rights and safeguard public values – in particular security, democracy, transparency, self-determination, non-discrimination, participation, privacy and inclusiveness. This is also necessary to be able to responsibly capitalize on the opportunities of the digital transition. Just like an innovative and level economic playing field with fair competition, consumer protection and broad social cooperation. When it comes to algorithms, legislation and oversight will ensure that algorithms and their applications are fair and transparent, so that citizens and businesses can rely on proper treatment.
human rights
At Van Huffelen’s request, the Rathenau Institute conducted research into the protection of human rights when using algorithms. The study identifies the measures implementing organizations take to safeguard human rights in the development and use of profiling systems and how they can strengthen this safeguard. The State Secretary endorses the recommendations of the research report and concludes that the government should base itself on clear legal requirements and ethical principles, which are unambiguously applied everywhere. For example, several guidelines have now been issued and methods have been developed to test algorithms for risk, but this requires more streamlining and prioritization. This should result in an implementation framework for the use of algorithms. It is also very important to initiate a discussion about the use of algorithms within the government and between governments and to share knowledge for a good implementation.
Van Huffelen: “Algorithms must be tested in advance and repeatedly examined during use for data quality, proportionality and possible bias. It is particularly important to focus in any case on algorithms with a high risk of violating fundamental rights. Clear supervision must be set up on these algorithms, before, during and after use. And the operation and deployment must be transparent, whereby it is clear where citizens and companies can obtain their rights if they believe that the algorithm is being used unlawfully.”
European legislation
European policy and legislation (in the making) play an important role in arriving at a responsible use of algorithms. In a European context, the government is strongly committed to better protection of citizens (including children) and to ensure that only algorithmic applications are put on the market that respect fundamental rights. The Netherlands is therefore committed to ensuring that the AI (Artificial Intelligence) regulation explicitly requires a risk analysis with regard to fundamental rights, that it must be made transparent where governments use AI and that there are legal remedies for citizens whose rights have been violated. Where necessary, we set additional requirements at national level. For example, we are going a step further for the Dutch government and are setting up an algorithm register and working with government authorities on the aforementioned implementation framework ‘deployment of algorithms’. The elaboration of the algorithm register and the implementation framework will take place next year in close collaboration with government authorities and experts. The State Secretary is expected to publish the Digitization work agenda in November of this year, containing an elaboration of the cabinet’s ambitions on this theme.