I recently came across this great cartoon from The New Yorker magazine by Paul Noth. Those of you who are familiar with the famous "Trolley Problem" (see also a great video clip from Harvard professor Michael Sandel, who introduces his course on "Justice: What's The Right Thing To Do?" with a discussion of the "Trolley Problem" ) will recognize the point immediately!
The "Trolley Problem" is a philosophical thought experiment first described in a 1967 article by Philippa Foot, an American philosopher (and, on an interesting side note, granddaughter of former US President Grover Cleveland). While Foot described the thought experiment, Judith Jarvis Thomson dubbed the experiment the "Trolley Problem" in a 1976 paper.
The thought experiment basically asks whether it is ethically acceptable to kill one person in order to save five people. In essence, the problem is usually set up as follows:
1. There is a runaway trolley moving down the tracks.
2. Up ahead, on the tracks, there are five people tied down to the track and unable to move. The trolley is heading straight towards the five people and will likely kill them.
3. You are standing some distance off in the train yard, next to a lever. If you pull this lever, the trolley will switch to a different set of tracks.
4. Up ahead on the side track, there is one person tied down to the track and unable to move.
The reader is left to ponder one of two choices - do nothing, in which case the trolley will kill the five people on the main track, or pull the lever, which will divert the trolley onto the side track, killing the one person (but saving the five).
The "Trolley Problem" has become very popular since Philippa Foot's first description in 1967. The television sitcom M*A*S*H, which was about a U.S. Army mobile army surgical hospital (which explains the abbreviation) during the Korean War and ran on CBS from 1972-1983 used a version of the problem in its series finale episode, "Goodbye, Farewell, and Amen", which aired on February 28, 1983. More recently, the problem also featured in episode 19 of the television show The Good Place (check out the video clip). It's been the subject of a number of articles and books, both in the popular press as well as the academic literature.
Kudos to The New Yorker for identifying one of the most frightening possible uses of artificial intelligence (AI). On the one hand, it doesn't seem so realistic that AI would ever make a decision involving the kind of ethical trade-off posed by the "Trolley Problem". On the other hand, I could envision several real-world scenarios that involve a similar ethical decision posed by the "Trolley Problem".
Two patients need an organ transplant. There is only one organ available for transplantation. Who gets it?
Alternatively, imagine a critically ill patient admitted to the intensive care unit (ICU) who is not expected to live. If the patient's organs could be harvested, five or six lives could be saved, assuming the patient could donate both kidneys, liver, heart, and lungs. Should the medical team continue to provide care for the patient, or should they withdraw support and proceed to harvest the patient's organs?
Alternatively, two patients need care in the ICU. There is only one available ICU bed left. Who gets the bed?
Imagine autonomous (driverless) vehicles! If the car encounters a sudden brake failure, should the computer force the car into a barrier in order to stop it, thereby risking the life of the passenger? Or should the computer force the car out of traffic and on to a sidewalk, risking the lives of pedestrians who may happen to be walking there?
When you consider all of the permutations of the "Trolley Problem", it's frightening to consider that one day in the not so distant future, it could be a computer, and not a human, making the decision! Should we allow that to happen? These are exactly the kinds of questions that will confront society as the use of AI continues to grow and expand.

