A few years ago, Amazon used artificial intelligence (AI) to find the best candidates for new positions. It resulted in only men being seen as worthy candidates, according to Reuters. The algorithms turned out to have been trained on almost only applications from men. Thus, no women were considered their own. A new report makes the Directorate for Children, Youth and Families (Bufdir) concerned about whether artificial intelligence could have consequences for gender equality. Artificial intelligence that saves lives More and more AI projects are being developed in the public sector in Norway. One of them is the project “AI decision support in medical emergency calls” at Helse Bergen. The goal here is for AI to make it easier for those who respond to medical emergency calls, and at best save lives. The way it works is that the artificial intelligence listens to the conversation. He then looks through the patient’s medical history and analyzes what is said. He will then be able to tell the 113 operator if a given condition is particularly dangerous, in the first instance stroke. But what if the caller does not speak Norwegian well? – An obvious challenge in our project is that callers to 113 speak several languages and dialects, says project employee Emil Iversen. Emil Iversen is a senior physician at the medical emergency telephone in Oslo and a researcher at KoKom at Haukeland University Hospital in Bergen. Photo: Erik Lerbæk / Erik Lerbæk The project is still early in the development phase. Nevertheless, they are already very aware of the danger of discrimination when using AI. Not everyone in the public sector is equally aware of this. Three per cent of the 491 public enterprises questioned believe that artificial intelligence can increase the risk of discrimination. This is shown in a new report prepared by Rambøll and Vestlandsforsking commissioned by Bufdir. – It is worrying that there is so little knowledge about equality and discrimination in artificial intelligence projects, says department director Anna Bjørshol in Bufdir. Anna Bjørshol in Bufdir believes it is important that more people become aware that AI can lead to the risk of discrimination and inequality. Photo: Ida Yasin Andersen/news – Experience from abroad shows that it can have major negative consequences for certain groups in society, as in the example from Amazon, says Bjørhol. Easy to solve AI expert Morten Irgens shares Bjørshol’s concern. – These systems seem so smart until I go a little deeper into them. We confuse them with knowledge when it is actually just text. The AI system is trained on lots of data from the internet and from other sources. If this data only contains one type of information, the AI system will not take into account the diversity in society. Irgens explains that. – Although the data only has information about or from white men in their 50s, the system does not provide any good advice that takes into account the diversity in society. Then the system can come up with advice that is discriminatory and sometimes prejudiced. Morten Irgens is project manager for research and artistic development work at Kristiania University College. He has been working with AI since the late 80s. Photo: Kristiania/Jonatan Quintero Irgens believes that it is easy to solve this problem if we just think a little differently about what AI is. – The users of these systems must be trained not to think of the systems as know-it-alls, but as eager learners who sometimes get it completely wrong.
ttn-69
Artificial intelligence creates concern for equality – news Culture and entertainment
