The debate on the ethical justifiability of artificial intelligence is currently focusing on the fact that machines shouldn't make unethical decisions that could reinforce existing prejudice. This concerns areas such as job applications, lending, court convictions and the facial recognition software used by the police and public institutions.
Less attention is paid to the question of how AI can influence people's moral integrity, says Nils Köbis, the Professor for the Human Understanding of Algorithms and Machines at the University of Duisburg-Essen. Together with Iyad Rahwan and Jean-François Bonnefon, he published an opinion piece in the «Los Angeles Times». People tend to behave unethically when they can justify it to others, when they see others making similar ethical sacrifices and when they can do so together with others. Social influence plays an important role in people's moral compass.
From a psychological point of view, AI can serve as an advisor, providing justification for breaking ethical rules, write Köbis et al. in the paper «Bad machines corrupt good morals». For example, AI systems could analyse sales pitches and suggest that deceiving customers would increase the chance of maximizing profits.
AI's role as an "enabler" is also worrying. Humans could work with AI to deliberately cause harm, as is the case with AI-generated media, especially "deep fakes". Studies show that people have too much confidence in their ability to recognise deep fakes, which makes them particularly vulnerable.
Erfahren Sie mehr über den moralischen Einfluss von KI am 20. Europäischen Trendtag, der am 13. März 2024 am Gottlieb Duttweiler Institut stattfinden wird. Dort wird Nils Köbis als Referent Einblicke in seine spannende Forschung geben.