Questions About Driverless Cars and the IT Infrastructure Behind Them
Where are we going? And when?
I just read an article about the inexorable march to a future filled with driverless cars. Based on past history, things seem to be moving forward as one might expect: There’s already a consortium of prospective vendors and competing standards proposals, so I think we can rest assured that an entire generation of young people who don’t understand the phrase, “Can I have the keys, please?” isn’t too far in the offing. Naturally, a lot of folks are excited, but I think this whole driverless movement raises some questions that might be overlooked in the standards development process.
First, just how “driverless” is “driverless?” When you think about it, this is a pretty complicated question. For example, is there any way to manually override or correct a mistake by the car’s driverless driver? The speed of the windshield wipers is an endless argument with the Mrs. — too fast or too slow. I assume The Borg of Star Trek fame controls it, but why does it care since it probably uses radar? Another question that comes to mind is: Does the onboard AI (aka the driverless driver) control the radio for the entire trip? How do you let it know that you need to stop to go to the bathroom? And don’t get me started on a question of utmost importance: Which window will it use at the local Chik-Fil-A to pay versus pick up your No. 3 eight-piece combo meal?
Another issue surrounding driverless cars is how it will impact car design. Sure, you make the payments, but if you’re not the one who drives it, does it matter what it looks like? Your vehicle may be able to go from 0 to 60 in under five seconds, but it’s not your foot on the accelerator. If you have a car that you don’t drive, does the old question regarding the type of car you drive have any relevance anymore — especially since a car smart enough to drive itself will probably be able to answer the question for anybody who might ask? This then begs the question as to whether the growth of driverless cars takes us back to the early days of the Model T, where everything on the road looks pretty much the same except for the number of seats and maybe the color — “Hey, there’s a four-door Spud”.
I think cars without drivers are going to cause some consternation for insurance companies. If we assume everyone will have a driverless car because their onboard systems allow them to operate in a superior mode to our current human-operated fleet of vehicles, then aren’t we saying they are “accident proof?” If that’s the case, do we need car insurance at all? Even if we still did have insurance, if you were in an accident, who would be at fault? Would you even have to get out of the car, or would your car’s computer exchange insurance information with the other vehicles and then flip them off as they drive away? Obviously, there are lot of details to be worked out before we see any mass market rollouts.
Certainly, there are benefits to having cars with no drivers. Old people, for example, will be able to retain their mobility without subjecting fellow drivers to the chaos and carnage they leave in their wake, but I think there is also some economic dislocation that should be expected. If everything on the road is automatically operated, that means you’re going to have more than a few truck and taxi drivers needing to pursue alternative career paths, leaving us with a paradox of driverless taxis with a guy in the passenger seat trying to run up the fare on unsuspecting tourists.
From a data center perspective, I think driverless cars portend a long period of continued growth, as the data processing and storage requirements to support even a single autonomous vehicle are staggering. Intel, for example, estimates a single car will generate 4,000 GB in a single hour. Since, by definition, mass transportation implies more than a single car, operating multiple hours a day, the volume of data generated and processed quickly becomes measured in terms of terabytes and petabytes. All this translates into data centers becoming more common than 7-Eleven’s across suburban landscapes. While I quiver in anticipation of this level of data center ubiquity, as a future passive passenger, I think there are few questions that need to be answered before I turn to my automated chauffeur and say, “Take me home, James.”