Daphne Odekerken (Utrecht University) develops one selection aid tool for assessing unreliable web shops. The AI system looks at properties of web shops that have previously been assessed as reliable or unreliable. If a new webshop scores better than a previous one that was classified as bona fide, we can also consider it reliable.
‘People should always be involved in complex decisions, it is important that AI supports people, not replaces them,’ says Odekerken, who will receive his PhD on January 6 for the design of an AI system that simplifies filing a complete tax return.
The police receive hundreds of online reports per day. The problem with this is that people often do not know what information is needed on the form to complete a tax return. Sometimes the police have to email back and forth to supplement missing information.
To make the process more efficient and reliable, Odekerken contributed to a selection aid tool, a smart declaration form. It is able to automatically identify missing information in the text that a citizen enters, and can ask specific questions.
The decision aid tool is based on rule-based AI. This means that you work with predetermined rules, such as articles of law.
Rule-based AI can only reason with complete information. And that is exactly what is lacking with the declarations, because they are usually incomplete.
Odekerken conducted her research within the National Police Lab AI, a collaboration between the police, Utrecht University and the universities of Amsterdam (UvA) and Delft (TU Delft). Within this Lab, the PhD students work simultaneously at both the police and the university.
Source: www.emerce.nl
