The collective eco-political swoon over the concept of electric cars and their self-driving capabilities is hitting a few unnerving speed bumps on the way to the promised land of an accident-free and nonpolluting transportation nirvana. 

Replacing the human intelligence, or lack thereof, behind the steering wheel will be a system of sensors conveying a myriad of bits and bytes to an all-wise on- board computer.  The output of this never tiring, never distracted, never emotional silicon brain is to provide society with a multitude of benefits which are said to include fewer accidents and fewer traffic jams.  Your silicon chauffer comes endowed with artificial intelligence (AI), a term which raises concerns for me (must be a luddite). Why does it have to be artificial intelligence?  Such a term evokes an immediate distrust of not having genuine intelligence at the wheel, a situation not unknown to parents of teenage drivers. I would rather that nothing artificial were in charge of my life as I am being driven down the highway or weaving through traffic jams.

By way of a progress report, a look back at where the incentive and push to deploy these “life savers” originated is instructive.  There was no spontaneous collective cry from the driving public demanding an autonomous car.  What began as a science-fiction notion in the 1930s was taken up by various automotive developers and commercial companies over the ensuing decades as is well-documented in a recent publication.  Insurance industry statisticians provided a compelling case for savings in car accident and medical payouts.  Urban planners saw the future in which a maze of closely spaced cars all moved to their destinations without accidents and with maximum efficiency in roadway use.  Better fuel usage and fewer vehicle emissions appealed to the environmentalists.  Government agencies envisioned a vehicle tracking system able to find any vehicle, anywhere, all the time.  Over the years, test vehicles by Mercedes, Toyota, and Google were able to fulfill their engineers’ goals of completing a series of self-driving exercises without incident in a controlled environment.  Arguments for and against autonomous driving cars are many.

However much these idealized visions of an autonomous car future may appeal to some,  their record of  recent traffic accidents and fatalities, including  a Tesla vehicle, indicate that fully autonomous cars are not yet ready for the open road.  Major automotive companies are taking a more cautious approach by offering bits and pieces of the technology, and labeling them as driver aids. Hazard warning, getting too close automatic braking, holding the car within its lane are now options for the new car buyer.  How well do these systems work?

An August 7, 2018 report, “Reality Check,”  from the Insurance Institute for Highway Safety (IIHS) illustrates the shortcomings when the latest  silicon  technology meets the road.  When subjected to real world driving control of a car, these current systems from top-tier automotive manufacturers all flunked basic driver’s education, in one manner or another.  Whether it was failure to keep the car centered in the lane, holding to a curving roadway, or stopping in time to avoid an object in the road, these systems demonstrated their need to go back to driving school.  Humans do a much better job of distinguishing a reflection of the open road ahead on a shiny truck body and stopping their car than a slow learning AI algorithm which drives into the open road image and truck.  However alert the AI driving aid may be, if it is not 100 per cent reliable, it becomes just another driving hazard.

If a driver must be alert to the unpredictable foibles of an AI-empowered car, he or she must be alert all the time… a state which should characterize all drivers anyway. The automotive conundrum presented by the new technology is a cross between Groucho Marx’s “You Bet Your Life” and a game of Russian Roulette. The AI intelligent car is a safer one, except when it isn’t, but there is no way to tell ahead of time.  How relaxing is it to relinquish your full attention to your computerized driving aids knowing that at any moment they may experience a misjudgment requiring your immediate attention to correct? The objective results of this IIHS report argue against any such reliance.  Those few drivers who actually read their car instruction manual will probably find that corporate lawyers have included exculpatory language which states that the driver takes fully responsibility for operating the vehicle at all times.  Nothing reliable has been gained by adding AI features, except an extra layer of complexity between the driver and control of the car.  Shut your eyes for a moment while driving and neither AI or nor your brain may prevent death.  Rather spend the money on continuing driver’s education, more frequent coffee stops, and smart phone restrictions for the driver.

No mention was made of the potential for malicious computer hacking of targeted AI cars, and the complete loss of driver and on-board computer control.  AI-controlled commercial trucks present a tempting target for a new generation of computer hackers and truck hijackers.

“Look ma, no hands” driving is a bit further down the road than advertised.

Charles G. Battig, M.S., M.D., Heartland Institute policy expert on environment; VA-Scientists and Engineers for Energy and Environment (VA-SEEE).  His website is www.climateis.com