Debate needed now on ethics and liability issues to get driverless cars back on Hong Kong roads

WATCH: Can Tesla’s new ‘autopilot’ system work in crowded Hong Kong? The South China Morning Post takes it for a test drive

In the event of an unavoidable collision, the decision-making of an autonomous car would ultimately rest in the hands of programmers writing the algorithms. Consider a revised version of the “trolley problem” used in ethics discussion: a child suddenly dashes onto the road, forcing the self-driving car to choose between hitting the child or swerving onto an oncoming truck in the other lane. With an autonomous car, the manufacturer now has the responsibility of planning such a scenario beforehand, as opposed to a person making a split-second decision. The algorithms, which decide how the car will behave, should be regulated to ensure it reacts responsibly. Jean-Francois Bonnefon at the Toulouse School of Economics in France suggests that the algorithms will need to accomplish three objectives: be consistent, not cause public outrage, and not discourage buyers.

Uniform algorithm standards could ensure consistency and predictability. However, this is difficult to implement since vehicles differ in size, acceleration and braking. Applying the above scenario, an SUV may choose to collide with the truck since it would offer better protection to occupants, while a compact car would probably choose not to.

Keeping it simple could avoid the ethical dilemma altogether. A project called CityMobil2 tests automated transit vehicles in various Italian cities. The vehicles simply follow a route and brake if something gets in the way.