The Uncertain Future of Self-Driving Cars

We keep hearing more and more about the wonderful self-driving car and how it is the future. Companies like Tesla, Google, GM, Waymo, and ZMP are announcing plans to produce fully autonomous vehicles. At first glance, these goals seem closer than ever. However, many AI experts express their doubts. We have already seen reports of multiple deadly accidents involving self-driving cars, most notably the incident in Arizona with one of Uber’s cars. Self-trained systems need to be able to adapt with the chaos of the real world, and it may take years before they can consistently avoid accidents.

Deep learning requires a vast amount of data and must be able to predict every possible scenario, however unlikely. The majority of the accidents documented so far were caused by unforseen circumstances. A few AI experts compare the push for self-driving cars to chat bots. The technology just isn’t there yet. However, some company executives argue that it is less about the technology and more about pedestrians. They suggest that pedestrians learn to anticipate the behavior of autonomous vehicles and abide by traffic laws.

Some companies have decided to move away from deep learning towards alternative AI techniques, but the techniques are deeply guarded and in their beginning stages. Despite the difficulties, Uber, Waymo, ZMP, and GM are moving confidently forward in the development of fully self-driving cars. Toyota jumped on the bandwagon and invested 500 million dollars in Uber’s self-driving initiative. Waymo has already begun testing their vehicles on public roads in Arizona. The Japanese company ZMP seems to be pushing ahead of its counterparts, debuting its self-driving taxis in the bustling city of Tokyo, with much success. ZMP aims to have the service be fully functional in time for the 2020 Summer Olympics. Most surprising of all, GM will begin the production of fully autonomous cars with no steering wheels in 2019.

Autonomous driving technology has quite a way’s to go, and for now, we will have to settle for semi-autonomous cars that still require some human intervention. It is very plausible that AI techniques will grow and improve over the next coming decades, but in the near future, companies potentially face what in the industry is known as an “AI Winter” and the financial difficulties that come along with it.