The Dutch government is expected to be one of the first to implement several important provisions of future EU legislation to regulate artificial intelligence (AI) as part of an ambitious digitization work programme.
In recent years, the Netherlands has been at the center of a far-reaching scandal over a flawed algorithm that has led to tax authorities falsely accusing thousands of people of child benefit fraud.
The Hague is now looking to lead the scrutiny of algorithms from a human rights perspective and the creation of a national database to track AI systems used by public bodies. It therefore anticipates two aspects of the European regulation on AI (AI law), historic legislation aimed at regulating AI applications based on their harmful potential.
“AI regulation should lead to explainable AI systems that respect human rights and lead to meaningful transparency that breeds trust, which is, of course, an essential component of a thriving AI market and is also important in a democracy “Dutch State Secretary for Digitization Alexandra van Huffelen told EURACTIV.
The work program adopted by the government also includes new initiatives to promote digital skills and a digital portfolio.
Public Registry and Supervision
From From 1 January 2023, a new watchdog will apply a set of new rules that anticipate future EU rules on transparency obligations, conformity marking (CE) and the right to complain.
Similarly, next month the Dutch government will introduce a mandatory register of AI algorithms used by public entities, from central government to local authorities. The rationale behind this decision is that, in order to actually benefit from the right of appeal, citizens need to know that an AI is being used.
In the initial version of the AI regulation, only providers of high-risk AI systems had to register in a European database. However, following pressure from, among others, the Netherlands, this obligation has been extended to public authorities using an AI-system.
One of the models of the Algorithm Transparency Register is the City of Amsterdam, which was established on a voluntary basis by three different groups: the general population, civil servants and technical experts.
“We have to work together with companies. Otherwise, we cannot make our algorithms transparent. IT departments are strongly interested in openness and dialogue”said a representative of the municipality of Amsterdam during an event in the European Parliament last week.
In addition, public authorities will need to assess the human rights impact of algorithms before using them.
To this end, the Dutch government has developed an AI impact assessment to assess the human rights risks of the device. If successful, the system will receive a mark of conformity.
“Evaluating artificial intelligence is not just about looking at the technology, but also at the context in which that technology is being used. We don’t want to blindly stare at a black box. It’s important to consider how this black box is being used.”said Mirko Schafer, associate professor at Utrecht University and co-author of the impact assessment, at the same event.
Schafer explained that the assessment is based on the context and added value of using the algorithms, how the models were trained, the risk mitigation measures in place, how often the AI needs to be reviewed, and the potential human rights implications.
The register is made public so that the supervisory bodies can judge the deliberative process, in which civil society organizations and journalists can also participate. The register will become mandatory for private companies once the AI regulation is adopted.
Van Huffelen believes that the Dutch impact assessment does not go beyond the EU Council’s position on the AI regulation, which she believes is quite strong.
“The Netherlands is committed to ensuring that this position is not watered down in future negotiations”she added.
The Dutch authorities are considering it 20 % of the population of the Netherlands does not have basic digital skills. This phenomenon is not limited to the ignorant fromit also affects those who have to undergo new training to acquire the digital skills necessary for the‘practice of current professions.
The agenda also includes new digital skills training programs to teach coding as early as elementary and middle school. New training offers will also support them at regional level, with the particular aim of training or retraining school leavers in advanced digital skills.
The Hague will also introduce digital information points in public libraries, where citizens can get help accessing essential public services online. In this regard, the Dutch government is also working on the creation of a digital wallet that anticipates future European regulations.
This wallet will include not only the national identity card but also all public documents such as driving licences, health insurance and diplomas. It will respect the data minimization principle according to which data must be limited to what is necessary in relation to the purposes for which they are processed. So if a store uses it to verify a person’s age, they won’t be able to access other information like their home address for example.
A beta version of the wallet is expected in the coming weeks, with the official launch scheduled for 2023.
[Édité par Anne-Sophie Gayet]