When algorithms become decision-makers

On behalf of the TA-SWISS Foundation, the ITA – together with the University of Zurich and Empa – investigated the opportunities and risks of artificial intelligence. For co-author Johann Čas of the ITA, transparency and compliance with basic civil rights are crucial.

"If, for example, we are confronted with artificial intelligence at the doctor's office or during our university studies, we should be distinctly informed about it", emphasizes Johann Čas, project manager at the Institute of Technology Assessment (ITA) of the Austrian Academy of Sciences. He and his colleagues presented the study "When Algorithms Decide for Us" at an online press conference on Wednesday. Among other things, the study states that "Citizens, consumers and employees should be informed as transparently as possible about the use of AI when dealing with the state, companies or their employers".

But will it work?

The study focuses on the use of AI in the areas of work, education and research, commerce, media and administration. Where can AI support human decisions? Can it even replace them? "AI must never be a substitute for human judgement", Čas is certain. His research at the ITA focuses on the ethical aspects of new technologies. "Ethics in the context of AI means, for example, ensuring that no one is discriminated against by AI. In Austria, there has recently been a public debate about the AMS algorithm. Even private recruitment agencies and companies are already using AI when it comes to identifying suitable candidates for a job".
So how can sufficient transparency be ensured? "First of all, customers, patients and other groups must have the right and the opportunity to view their data and, if necessary, to correct entries accordingly. They should also have the right to put decisions made by AI into question," Čas is convinced.

AI as a job killer?

The experts also see the much-feared use of AI in different fields of labour as an opportunity: the productivity gains that AI can bring about could be used to shorten working hours and to distribute work and the income associated with it more fairly; new services and products could also open up new fields of work. Caution is called for when it comes to its use in the media: AI is already being used, for example, to create and distribute fake news; the media and the state would therefore also have the task of identifying and disproving fake news. (DR)

The study is available in German as an Open Access eBook:
Wenn Algorithmen für uns entscheiden: Chancen und Risiken der künstlichen Intelligenz