TEQnation 2021 – Lisa Noorlander & Ramon Ankersmit – AI, How do we deal with it?

In the present age of information technology, many data processing machines govern our daily lives: from servers that keep records of our financial transactions to mobile phones that track our voices and locations. Every day these machines become more independent and intelligent. Using artificial intelligence techniques, organisations and governments can lighten their employees’ workloads, find errors, and predict results. At the Dutch tax office, several departments already use these techniques to improve their methods and processes. No matter how great this may sound, we must remember that these AI techniques are used both for people and by people. And people simply make mistakes. So, do AI models actually respect our human rights? Do we have any privacy left? Can we assume AI makes decisions free from biases? During this session we explain how to make a reliable AI model. We also show what products we are already developing using machine learning at the Dutch tax office and how we include ethical considerations in our development process. Lastly, we will show what happens when we let the free Internet loose on AI without caution…