Following some rather high-profile problems with supposedly “self-driving” Teslas, automotive industry analysts are once again raising issues about the future of this technology, though Elon Musk seems to doggedly believe that it’s just around the corner. Let’s hope that the corner in question is well-lit with clear stripes painted on the road. After one Tesla decided to slam into a tree at high speed killing both of the vehicle’s occupants, alarm bells were going off. More questions were raised regarding another Tesla that kept slamming on the brakes every time it passed a billboard with a stop sign painted on it. Musk had previously claimed that we would all be riding around in driverless taxis by 2020, but at this point, we should be asking if this technology is really ready for prime time. Or is it even close? Some of our technical wizards believe that fully autonomous vehicles are still decades away, assuming we ever have them at all. We might not, and it can all apparently be blamed on the fact that the artificial intelligence just isn’t intelligent enough yet. (Wall Street Journal, subscription required)
In contrast to investors and CEOs, academics who study artificial intelligence, systems engineering and autonomous technologies have long said that creating a fully self-driving automobile would take many years, perhaps decades. Now some are going further, saying that despite investments already topping $80 billion, we may never get the self-driving cars we were promised. At least not without major breakthroughs in AI, which almost no one is predicting will arrive anytime soon—or a complete redesign of our cities.
Even those who have hyped this technology most—in 2019 Mr. Musk doubled down on previous predictions, and said that autonomous Tesla robotaxis would debut by 2020—are beginning to admit publicly that naysaying experts may have a point.
The first thing to point out from this report is that there are currently no true “self-driving cars” in operation aside from a few beta-testing models. And even those only work reliably on controlled, ideal courses during perfect weather conditions. The vision we were originally pitched of a “self-driving car” implied that you would soon be able to hop into the back seat, tell the vehicle where you were going and it would take you there. There are no cars on the market that can do anything close to that.
The term “self-driving” has been dumbed down quite a bit in recent years by many industry influencers, including Elon Musk. As I pointed out after the recent Tesla crash, even the vaunted Tesla Model S isn’t close to being truly autonomous. It only works that way when the driver is in the seat and has their hands on the wheel at least every ten seconds. You always have to be alert and ready to take control in case the car messes up. That doesn’t free up any of your time and just puts a much larger hole in your bank account for the privilege.
Computer scientists contacted by the WSJ explained that the complexities involved in driving a vehicle through all of the conditions it might encounter make it a task that is far beyond the capabilities of current technology. In order to have a machine replicate all of the abilities of a human driver, you would need an artificial intelligence system that’s at least as good as a human being’s brain, decision-making abilities, and reaction times. “For a car to drive like a human, researchers have to create AI on par with one.”
They went on to say that not only are we not there yet in the field of AI research, but we don’t even know how to get there. Other major breakthroughs in AI would have to be made. And then we’re talking about something that’s dangerously close to “General AI,” or a true, self-aware robotic system. And I’m pretty sure we all know how that movie ends.
Driving your car to the grocery store and back is actually far more complicated than putting a rocket into space. The rocket only has to travel in a relatively straight line up and down through empty air. And even then, launches are regularly scrubbed if unexpected clouds move into the area. Meanwhile, when you drive a few miles through the city to pick up some food, you are constantly monitoring other vehicles for unexpected maneuvers, pedestrians and bicyclists on the side of the street, shifting road conditions, detours, and who knows what else? You’re making hundreds, if not thousands of decisions on every journey even if you do most of that work subconsciously.
We don’t have Artificial Intelligence today that can function at that high of a level of instant data collection and decision-making. And some of the top people in that field still don’t know if we ever will. So I’m sorry to say it, Mr. Musk, but the true self-driving cars we were all imagining are likely still the subjects of science fiction, at least for now.