The Moral Machine Can We Teach Artificial Intelligence Right from Wrong?

Imagine you are sitting in the passenger seat of a self-driving car. You are moving at highway speed through a tunnel when suddenly a group of pedestrians appears in the road ahead. The brakes fail. The car's onboard computer has exactly one option left: swerve. But swerving means hitting a concrete barrier, which will almost certainly kill you. Staying the course means hitting the pedestrians. The car must decide. And it must decide in milliseconds.

Here is the question that keeps philosophers, engineers, and policymakers awake at night: who programmed the answer? Whose values are encoded in that split-second calculation? And more fundamentally — can values even be programmed at all?

This is not a thought experiment confined to university seminar rooms. It is the lived reality of a world in which artificial intelligence is being handed the controls to some of the most consequential decisions human beings have ever faced. From highways where autonomous vehicles share lanes with cyclists and schoolchildren, to emergency rooms where algorithms help doctors decide who receives a ventilator and who does not, to courtrooms where risk-scoring tools shape sentences before a judge has spoken a word — AI systems are making moral judgments right now, at enormous scale, often invisibly, and almost always without the informed consent of the people most affected.

What you are about to hear is the story of how philosophy left the classroom and entered silicon. It is the story of trolley problems on real roads, of ancient ethical frameworks being stress-tested by modern code, and of a remarkable global experiment that revealed just how differently human beings think about right and wrong depending on where they happen to have been born. It is, ultimately, the story of one of the most profound challenges of our age: teaching right from wrong to something that has never been alive.