The Driverless Car – Killer App for Artificial Intelligence?

Whilst the Driverless Car is yet to own the roads there are many examples of the use of Artificial Intelligence in technology products dating back to as early as 1947 when a C4 Aircraft was piloted by a punch card driven computer to accomplished the first entirely automatic flight across the Atlantic Ocean, flying from Newfoundland to England†.

Autopilots

A reporter for Time magazine narrated the event,

The plane behaved as if an invisible crew were working her controls. … The commanding robot was a snarl of electronic equipment affectionately known as “the Brain.” Everything it did on the long flight was “preset” before the start. It received radio signals from a U.S. Coast Guard cutter. Later it picked up a beam from Droitwich, England, and followed that for a while. When the plane neared Brize Norton, the wide-awake Brain concentrated on a special landing beam from an R.A.F. radio and made a conventional automatic landing. On the way over, the crew checked the course and watched the instruments. Most of them had little to do. They played cards and read books. ‡

There are many other significantly more modern examples including the Deep Blue playing chess, Watson besting humans at Jeopardy and most recently Google’s AlphaGo beating a Grand Master (Lee Sedol) at Go. There are other less visible examples such as the engines under-lying Google translate which uses Deep Neural Nets and the algorithms being used to automatically generated sports commentaries of amateur sports from the results posted online.  However, is there a Killer App for AI – “a feature, function, or application of a new technology or product which is presented as virtually indispensable”?

AlphaGo

AlphaGo Results

I would argue that in the minds of the public such an application for Artificial Intelligence does not yet exist.  There is no capability that is so indispensable for our daily lives that if it ceased to exist or function our lives would be dramatically different. More importantly there is no such function that is inextricably linked with Artificial Intelligence in the the minds of the public.  Or is there?

It could be argued that flying would remain inconceivably dangerous without the sophisticated array of automation and Artificial Intelligence used in modern Passenger Aircraft.  Humans simply could not fly these machines without this assistance, and indeed the Aircraft can fly themselves without requiring human pilots. The Airline industry is quite deliberate in ensuring that the flying public believe that there is a human at the controls of the Aircraft.

The Driverless Car

This brings us to the Driverless Car.  There are many definitions of these autonomous vehicles but essentially we are talking about a vehicle that can operate on the public road system without the intervention of a human driver.  This will undoubtedly be a “Killer App” if not the “Killer App” for Artificial Intelligence and robotics.  A Driverless Car is a robot, just not one with an android or any other biologically inspired form factor. The have all the necessary components to make them a fully functional Artificially Intelligent robot.  The will have situational awareness, multiple senses including Stereoscopic vision, LIDAR, Radar, Ultrasonics, Wireless, Voice and GPS and the ability to act out their intentions.  The will interact intelligently with their passengers and each other and other non-AI based road users including pedestrians and cyclists. They will perform a vital function in a much safer and more economical fashion than human drivers ever could.

For example, every human driver needs to go through a training program and ultimately a steep learning curve with the inevitable crashes occurring.  This is directly reflected in the young driver insurances premiums as they tend have more frequent and more serious crashes.  With the driverless car, every vehicle has the most experienced driver capability possible (assuming their system is kept up to date) and the experience of every Driverless Car can be shared with every other Driverless Car. This means that when a novel circumstance is encountered that experience can be share with every other vehicle, cheaply and automatically.  The number of injuries, fatalities and cost associated with inexperienced human drivers will be virtually eradicated saving perhaps hundreds of thousands of lives and many billions of dollars in medical, legal and repair costs annually.

Additionally it will not be possible for a Driverless Car to violate a traffic rule unlike human drivers. Humans generally drive in a manner that skirts the boundaries of the road rules constantly and some individuals drive recklessly by speeding or driving under the influence of drugs or alcohol.  Crashes caused by this criminal behaviour will not exist. Furthermore, crashes caused by fatigue, distraction or poor and impaired judgement will be a thing of the past.

Of course it will take some time and significantly more technical, legal and social advances before this vision can come to pass and there will be crashes and failures cause by manufacturing and design defects in the vehicles, their systems and the Artificial Intelligences controlling them but these will only every be once off events. The lessons learned can be shared and embedded in the global fleet of Driverless Cars unlike humans who are doomed to repeat the same mistakes over and over again as each one “learns” to become a safe driver.

 


 

† (Source: Praise the Machine! Punish the Human! The Contradictory History of Accountability in Automated Aviation, M. C. Elish and Tim Hwang Comparative Studies in Intelligent Systems – Working Paper #1 V2 Intelligence and Autonomy Initiative1 — May 18 2015, Data & Society Research Institute).

‡ (Source: 5 “No Hands.” 1947. Time 50(14): 63.)

Comments are closed.