Sep 21, 2016

Of Paradoxes and Human Errors: Implications of Driverless Cars

Ashish Kumar Jha

An accident involving Tesla’s much coveted autopilot programmed car has given rise to a host of issues around driverless cars. Imagine a future where hordes of fully autonomous cars take over our streets. In that same future, a car cruising at 150 miles per hour develops a technical snag, reboots automatically and in the melee crashes with another car resulting in unfortunate deaths. The question now is who is responsible for the death. Does it lie within the ambit of industrial accident and hence no individual goes punished. Is it the fault of the owner of the faulty car or the engineer who programmed the car? Can we lay the blame on the doorsteps of the quality engineer who certified the car?

All these questions force us to look forward into the evolving relationships between man and machine. Humans by nature seek retribution. The blame has to lie on someone, someone identifiable and accountable. Numerous works of research have shown that our uneasiness with machines does not lie in trusting their efficacy or accuracy. Even the worst critics of automation cede that machines even on their worst days outperform humans. The issue that makes us uncomfortable with machines is the lack of accountability. What if something wrong happens, who will be accountable for such a condition. The doomsday war between humans and machines circa the “Terminator” movie series is going to emerge, if at all, due to the lack of any human accountability in machine related mishaps.

The life of a proponent of the driverless car has not been easy in recent years. The issue raised above is in addition to the multiple ethical and moral dilemmas already surrounding the driverless cars. The moral questions regarding “who to kill” in the event of a certain accident has occupied popular media over the past few years. There is no easy or widely acceptable answer to these questions. Humans prefer social well-being over individual when they are not the individual in question. The answer to “should a car kill the driver to save lives of 10 pedestrians” and “would you sit in a car which is programmed to kill you to save others” is often not the same. I would not go into the swamp of discussion on the morals of deciding a pecking order of human lives.

However, these moral dilemmas deserve some thought because whether we like it or not driverless cars are fast becoming a reality. Even if we leave the questions of the pecking order of human lives there is a second concern. Who decides the pecking order? Do we let free market competitive firms decide the pecking order and risk falling into the traps of ever more secure cars for passengers? The cars could evolve on to become killing machines on streets. Or we let government regulations do the job. After all government already does a lot of regulation making in the name of social welfare. The risk there is that very few people would willingly buy a socially relevant car designed to sacrifice them.

But all the moral dilemma moments rely on one basic assumption that the computer in the car does not have enough time to apply brakes and stop the accident or fatalities but has enough time to compute the pecking order of deaths and make a choice of the sacrificial lamb. What if the computer chip is slow and the well accepted pecking order of lives could not be computed. What would the repercussions of some non-computed lives being lost be? This goes back to the first point that I rose “who is to blame”. How similar are the issues of slow computer chip and human with slow reaction time. How similar would the situation of a chip not clearing its cache and hence becoming slow be to the situation of drunk driving?

There are no easy answers to these questions but our ability as a society to integrate with the AI enabled driverless cars of the future rests on us discussing these issues. We need to better understand the issues surrounding the ethics and morality of the cost of human life. We need to dig deeper into our own souls to rise above our instinct of seeking retribution for loss of life. The integration of these systems into the society depends dearly on the quality of discussion on these topics.

AppLy Now