The recent Ethiopian airline crash that claimed the lives of all 157 passengers on board is suspected to have been due to a software glitch on the Boeing 737 Max 8 aircraft. The latter is a new model from the Boeing stable that uses an automated flight software called the Manoeuvring Characteristics Augmentation System. This automatically lowers the nose of the plane when it receives information from sensors that the aircraft is flying too slowly or steeply. However, in the Lion Air crash in Indonesia last October where 189 people on board the same model of aircraft died, the automated software erroneously pulled the plane’s nose down at least 20 times as the pilots struggled to manually override the system. A similar issue is believed to have plagued the Ethiopian carrier as well.
There’s no denying that advancing technology has made aviation the safest mode of transportation in the world. Yet, as recent air crashes show, technology is a double-edged sword. As the aviation industry moves increasingly towards automated systems driven by a desire to eliminate human error, it seems to account less for technology error. This raises a fundamental debate – are we handing over too much power to technology?
With the advent of artificial intelligence, society is already in the throes of great change. Innovations ranging from self-driven cars to Internet of Things are based on the belief that machines are more reliable than humans. But given that machines are input-driven, wrong inputs produce wrong results. Recall that in 1980 the US and USSR almost engaged in all-out nuclear war due to a false warning produced by a defective chip in American military computers. Greater catastrophes could befall us today if we allow AI to make all decisions without manual override. AI may be today’s technological reality. But we must find the right balance in the human-AI interface.