From Brian Horton, West Launceston, Tasmania, Australia
Clare Wilson reports a “real life” test of the “trolley problem”, in which subjects could allow five mice to receive a painful electric shock, or press a button to shock just one mouse (19 May, p 14).
As in all cases of the trolley problem, the situation is so artificial that people try to think of ways of avoiding the dilemma without actively harming anyone (or any mouse).
In the published study we see that the subjects found the test hard to turn down because they received course credits for it. One out of 198 people refused to take part but was still given credits.
Nearly half did not really believe that any shocks would be given – and no mice were harmed in this study. The only conclusion I can reach is that some people would behave differently in a real situation from what they say they would do hypothetically.
Advertisement
This will not help artificially intelligent cars decide how many people to kill (and which) when an accident is inevitable.
