Seven Ways Driverless Cars Could Fail
In my last post, 7 Wonders of the Driverless Future, I listed seven magical societal benefits that driverless cars might deliver.
While I am optimistic about driverless cars, skepticism abounds about whether driverless cars can actually work and, even if they are technically possible, whether they should be allowed.
In the interest of facilitating a robust conversation, I want to offer seven of the most pressing concerns. Effective decision making, as Peter Drucker observed, depends on the dialogue between different points of view and a choice between different judgments.
The following differing points of view and judgments cover a wide range of potential issues, including technical viability, operational resiliency, and even fears that driverless cars might work too well and thus unleash a host of unwanted secondary effects.
Please feel free to reiterate, refute or add to the items on this list.
1. There are too many “corner cases.” While driverless cars have demonstrated remarkable proficiency in many aspects of driving, some contend that many driving scenarios will remain unsolved. MIT’s John Leonard, for example, points to left hand turns into heavy traffic, adverse weather, changes to road surfaces, and eye and hand gestures as open technical questions for which solutions “might be a very long way away.” Leonard should know, he has been a central figure in the field for years, including a stint as the leader of MIT’s autonomous car team in the famed 2007 DARPA Urban Challenge. (Watch this excellent presentation for Leonard’s detailed assessment.)
2. The technology is not robust enough. Even if theoretical solutions are found, it might be hard to reliably translate them to the real world. Robustness would depend on a myriad of electronic, mechanical and software components operating with little tolerance for error under adverse (freezing, wet, boiling) conditions, long duty cycles, less than perfect maintenance (car owners being human), fender benders, network and electrical outages, and so on. Dr. Louise Cummings, for example, made a well-reasoned argument along these lines during a recent U.S. Senate committee hearing.
3. They will cost too much. The cost of sophisticated electronics, including lidar, radar, sensors, cameras, computing and networking devices, on top of the cost of additional development, maintenance, and liability, will price driverless cars out of reach of most consumers.
4. They will not mix well with human drivers. Even if driverless cars can learn to interact with human-driven cars, human drivers will not be able to deal with driverless cars. The resulting confusion would lead to more accidents and congestion, rather than less.
5. Regulatory and liability hurdles will long delay the technology. Legal and regulatory frameworks are all built around the assumption of human drivers. This encompasses expectations about who is liable for what and when, what components are required in cars, and so on. Sorting through the maze of necessary changes will take decades, and will offer ample points for entrenched interests to delay the approval process. Product liability might be a showstopper.
6. They will be hacked. Beyond the unintentional vagaries of the real world, driverless cars will suffer from attacks by hackers, hooligans, hijackers, thieves and terrorists. Risk range from invasion of privacy to the specter of driverless cars being used as precision bomb delivery vehicle.
7. They will cause economic devastation. By some estimates, 10% of all jobs are driving related. Driverless vehicles could put millions of Americans out of work, including taxi, Uber, bus and truck drivers. And, the list of jobs at risk do that stop there. Add auto suppliers, new car dealers, collision and repair shops, tow truck operators, insurance agents, adjustors, call-center operators, ambulance drivers, emergency medical personnel, traffic police, and untold other jobs related to how cars and trucks are designed, manufactured, sold and operated today.
These are serious issues and mandate that we approach the deployment of driverless cars with caution and deliberation.
Some issues embody technical challenges that developers must adequately address. Policy makers and regulators will have much to say, too, about acceptable standards and levels of verification.
Other issues, such as cost and user acceptance, will be decided in the market place. Driverless technology will enable tremendous business and service-level innovation. Customers, however, will have the last word on whether the products are good enough.
The hardest issues will be those that require making the societal tradeoffs between wonderful benefits and tremendous cost.
As with every major technological transition, driverless cars will result in winners and losers. There can be no neutral decisions. Whatever decisions are made (or not made) will result in gain to some and loss to others.
If you could choose, how would you decide?
This article originally appeared on Forbes.