A group of Japanese scientists decided to find out how people relate to moral judgments of artificial intelligence. The results of the work published by the magazine scientific reports , it was surprised: it turns out that we are more willing to agree when we agree when He gives something a positive assessment, especially if it contradicts the negative opinion of a person.
The study was led by the professor of the University of Rissue Hitoshi Yamamoto and professor at the University of Tsuda Takakhis Suzuki. Researchers considered a complex moral dilemma: is it worth helping people with a bad reputation. Such situations are often at a dead end, so they are ideal in order to figure out how we make decisions.
Two experiments with the participation of volunteers were conducted. Work scenarios were based on: the participants showed hypothetical situations where it was necessary to evaluate the decisions of two managers – a person and a computer program. In each case, the managers had to decide whether it was worth working with the employee, based on his reputation.
An interesting pattern has been discovered: if you need to decide whether to cooperate with a person, relying on his reputation, people really believe more than the judgments of artificial intelligence. This feature was especially pronounced when the algorithm justified the refusal to help, and the person condemned this behavior.
Researchers believe that people consider artificial intelligence a more impartial judge – unlike a person whose decisions can be influenced by prejudices and hidden motives. This conclusion is especially important now, when AI penetrates deeper into our daily life, and even into the sphere of justice.
Until now, scientists have observed two opposite phenomena in the interaction of people with AI: some avoid algorithms and do not believe their decisions (“algorithmic rejection”), others unconditionally take any machine verdict (“algorithmic admiration”). A new study sheds light on the nature of these phenomena and explains how doubts about the moral assessment of the situation affect the readiness of people to trust the machine.
The results of the study will be a support for developers of AI systems when creating algorithms that can make balanced decisions taking into account all circumstances. Understanding what determines our attitude to AI will help create technologies that harmoniously fit into public life.