
The Massachusetts Technological Institute launched an Internet project called Moral Machine . This is a site with a set of tests that should help scientists figure out how artificial intelligence should make a moral choice.
Moral Machine offers everyone to pass a test for a moral dilemma. If the car on autopilot gets into a situation where someone must inevitably die-passengers or pedestrians, what choice should it make? How to choose less from two evils - to kill two passengers or five pedestrians? As an external observer, the user must evaluate which decision is more acceptable. After passing the test, you can compare your answer with the answers of other people.
The launch of the site was preceded by the MIT study , in which scientists tried to solve the problem for what behavior it is necessary to program autopilot in such emergency situations. People surveyed for research revealed an ambiguous assessment of autopilot actions.
The respondents approved the choice of a car that left a collision with pedestrians and donated passengers, but they themselves were not ready to be in the role of donated passengers.
This summer, the first accident with a fatal outcome with the participation of a car on autopilot occurred.