If you're looking forward to a car that drives itself, you might have to put that dream on hold for a while.
Google's self-driving car has been a media darling, but Slate points out that the car faces many, many obstacles in addition to the legal and infrastructure issues autonomous vehicles already contend with. The big problem? The Google car just isn't smart enough yet.
It can't avoid potholes or deal with rain and snow, and it also relies heavily on extensive mapping programs that mark every stationary object. These maps are many times more complicated than Google Maps or Google Earth, and the car is absolutely dependent on them.
This drawback means all four million miles of public American roadway must be mapped, along with every driveway, trail and private road. Only a few thousand miles have been mapped thus far. Sure, Google says its test cars have logged 700,000 miles, but that's all been on the same roads, according to Slate.
And of course, maps need to be constantly updated. Signs and stoplights get changed or removed all the time, and if the Google car comes across an intersection that has had a stoplight installed where there wasn't one before, it might run a red light since it wouldn't know to stop.
Parking is an issue, too, since the car can't find spaces in retail lots or parking garages. Coned-off construction areas confuse it, too, and glare from the sun can cause it to misread a stoplight. It also drives around any obstacle it detects in the road--even harmless paper it could easily drive over.
At the end of the day, the problem is simple: No artificial intelligence system yet developed can reason as well as humans do. It may exist someday, but for now, increased driver aid technology is as close as we're going to get.