Governments are increasingly deploying artificial intelligence (AI) to better regulate traffic, for example, or to tackle complex social issues such as energy transition, poverty policy or climate change. By analyzing large amounts of data, new information becomes available on which to base policy decisions. This offers opportunities and possibilities, but there are also risks. CBS is actively involved in the developments surrounding AI, both nationally and internationally.
AI is everywhere. In the public sphere, AI is increasingly determining the pace and quality of new developments. With AI, processes can be accelerated and improved. It can also be used to discover patterns in data, which can lead to greater insight into complex issues. But there is also a downside. Barteld Braaksma, innovation manager at CBS, explains how delicate digital processes can be. 'With AI, it is crucial that the data are correct, that the training set - the algorithm's 'teaching package' - is set up transparently and neutrally, and that the outcomes are checked for any errors. When large amounts of data are involved, and somewhere a small error creeps into the algorithm's learning process, it can have major consequences for the outcome. What matters is the explainability and transparency of the system. In addition, the training data must be representative to get a fair algorithm.'
The development of the global Internet has long been the domain of private entrepreneurs. For a number of years there has been a growing awareness that the digital technosphere is also a public space, in which there is a need for regulation by means of laws and regulations. There is increasing attention and money for this from the government. For example, the Growth Fund is investing in the Dutch AI Coalition to make targeted use of the opportunities and possibilities of AI for Dutch society and the economy. Braaksma: "We are working in the Dutch AI Coalition with, among others, the University of Amsterdam, Utrecht Data School, TU Delft and the VNG on a digital toolbox to create neutral and fair algorithms. In addition, CBS works internationally with a large number of European partners in various networks and consortia such as TAILOR.'
Investments from the Growth Fund are being made in the Dutch AI Coalition to target the opportunities of AI for society and the economy
TAILOR is a European collaboration of governments, knowledge institutes and companies with 55 partners from all over Europe. With a joint research and development program, TAILOR aims to create a new standard for reliable and fair AI, where human values are and remain the starting point. Europe thus wants to play a global pioneering role in the development of responsible, humane digital technology. TNO is also a partner in the TAILOR consortium. Freek Bomhof of TNO explains: 'The developments in AI technology follow each other in rapid succession, and on that technological front Europe is participating, but not leading. It is, however, in research and initiatives that deal with the ethical application of AI. The need for legal frameworks that protect users is more of a political priority here. That previously resulted in the GDPR (General Data Protection Regulation, implemented in the Netherlands with the AVG, ed.). The AI Act, which the European Commission is now working on is another such step forward.' But it's not just about laws and regulations.
Awareness and some basic knowledge about AI among the general public is also essential for an ethical framework. "For that, we really don't need to train everyone to be programmers," Bomhof says. And he is also not afraid of dystopian scenarios in which people lose out to technology. 'Compare it to road traffic. When New York was in danger of disappearing under the excrement of horses due to the massive amounts of horse and carriage traffic, a technical innovation appeared that prevented that problem: the automobile. The disadvantages associated with this means of transportation - such as air pollution - were not known then. That basic knowledge is there now, and it is important for making choices about the environment and climate, for example. Similarly, it is important that users of AI - and we all are - have some understanding of how AI works in order to oversee the consequences of its use. In the Surcharge Affair, it wasn't just about the algorithm either. There were also people at the controls all the time. So it is about the technology, but also about the business processes themselves.'
In late 2021, TAILOR, together with research platform CLAIRE and the VISION project, organized a workshop on the use of AI in the public sector. The goal was to organize a broad conversation, with room for all the different perspectives. In that setup, the initiative succeeded, Braaksma and Bomhof think. ´Working towards a transparent and reliable AI is also a matter of demystification,´ says Bomhof. 'With an algorithm registry, with an open infrastructure for SMEs in all European member states, we are slowly but surely building a reliable digital public space. In that process, you can strongly see the European signature of AI in the public domain: the diversity and complexity of the continent, with more emphasis on human values. The importance of the individual is central to that. These are the values with which we are carving out the "roadmap trustworthy AI. It's not always the easiest path, but ultimately it delivers the most for society.'
