'AI as a tool for municipalities' is a digital magazine from VNG, CBS, the City of Amsterdam and Code for NL. It explains in understandable language what Artificial Intelligence is and what it can do for municipalities. The publication highlights the complex subject from multiple angles, from practical examples to the angels of algorithms that can increase and reinforce existing biases.
Author: Karina Meerman
The umbrella term AI or artificial intelligence covers a wide range of applications. The automatic completion of online searches by a search engine is a form of artificial intelligence, as is the route planner that continuously (re)calculates travel time on the basis of current information while driving. At the root of this are algorithms. These are instructions that should lead to a certain goal, like following a recipe should lead to a bowl of muffins. This is better explained by Maranke Wieringa of the Utrecht Data School (Utrecht University) in the article "What is an algorithm? She discusses the terminology and explains the distinction between narrow and broad AI. "Narrow AI is focused on a single task. General AI involves broader applicability. Then an application can, for example - just as humans can - play chess, go and lingo. However, that is still difficult to realize at present." Therefore, she and her colleague Mirko Tobias Schäfer recommend "calling a spade a spade" and specifying generic "AI" with words like "machine learning" or "image recognition.
AI in all its forms is driven by algorithms. There is also a growing awareness among municipalities that algorithms are not neutral or fair and may have a bias (a bias) in them because they are created by people who bring their own worldview into the creation of algorithms. This can lead to discrimination that should be avoided as much as possible. In the article "Fair Algorithm," "bias" is explained by Gerhard Dekker, head of Data Services and Alliances at CBS. He also talks about the tools CBS has developed to prevent discrimination based on algorithms.
The Utrecht Data School's free tool DEDA (The Ethical Data Assistant) ties in nicely with this. 'Grab Your Responsibility' explains how DEDA helps map ethical aspects of data projects and document the consideration process. The goal is to provide clear accountability to stakeholders and the general public.
Peter-Paul Verbeek also discusses the ethical aspects of AI. Verbeek is professor of Philosophy of Man and Technology and scientific co-director of the University of Twente's DesignLab. He himself answers his favorite question, "How people can flourish with the technology to be deployed and what are the conditions for that.
Tamas Erkelens, one of the initiators of the AI with Impact project and public tech lead at the municipality of Amsterdam, bridges the gap between theory and practice. He talks about the strategic alliance of CBS and VNG to develop better government services based on transparency and honest algorithms. He explains that together with CBS, he came up with the idea of (further) developing a number of open source solutions, so that they could then be shared with all other municipalities. "It would be a shame if they all started inventing the wheel on their own. We thought it made more sense for experts and the forerunners, including Amsterdam, Rotterdam and a few other cities, to share their knowledge in this area with the rest of the Netherlands."
Some of these examples are also featured in the magazine AI with Impact. Like Gem, the chatbot of the municipality of Dongen, a new way of communicating added as an extra service for citizens. Paul Geurts, strategic information advisor at the municipality of Nijmegen, explains how smart cameras count passers-by in the city center. The goal is more insight into the development of economic activity and (traffic) safety in the city. The municipality of Amsterdam uses image recognition (object detection) to detect garbage to get the city clean faster. The municipality of Nissewaard uses AI for clues to welfare fraud.
Source: VNG Realization Magazine
