Many people have dreamed about autonomous vehicles for a long time, with hopefuls imagining cities full of driverless vehicles that are not only safe but make commutes comfortable. But despite the excitement lasting over a decade, the progress appears minimal. In fact, many believe that despite claims that it is just around the corner, the pursuit of self-driving cars is going nowhere, even after around $100 billion has been poured into it.
The technology has attracted various companies over the years, including Google, GM, Ford, Tesla, and a massive number of startups looking to make a breakthrough. Accompanying them have been demos and promises, with plenty of confidence that things would change with one more step. Some companies have even insisted that they were already there, but the claims have never proven true.
Instead, the attempts at making the technology work have been unsuccessful, and in some cases, downright amusing. Robot cars, such as robo-taxis, struggle with basic obstacles. They are not capable of dealing with difficult weather patterns, and construction, animals, and even “unprotected left turns” prove serious problems.
While some people throw out “deep learning” as a solution, it is more memorization than actual learning. It might help these self-driving cars deal with one situation, but it might not apply to a very similar but different situation.
Birds in the middle of the street are a good example. Self-driving cars do not know that the birds will fly away as they approach, nor that other drivers know they will scatter. Therefore, an autonomous vehicle is likely to come to a stop, which could cause a traffic jam or an accident.
Moreover, birds are one of many “edge cases” that AI is not equipped to handle. It has the potential to turn ordinary situations on the road into something worse. That is part of the reason self-driving companies have taken to shortcuts, running simulations inside data centers and using that data to pad mileage and safety statistics.
George Hotz, the founder of startup Comma.ai, said, “It’s a scam.” He commented about how self-driving companies “squandered tens of billions of dollars,” and it does seem that way when you look at the numbers.
According to a McKinsey & Co. report, investors have spent around $100 billion on the field, and all there is to show for it are demos. Well, that and amusing videos on the internet depicting self-driving cars getting confused in ordinary traffic. On the side of fairness, there has been improvement, but it is not visible, and it is nowhere close to the “just around the corner” enthusiasm of many big players.
It does not help that humans are incredible drivers, which Hotz also spoke about. After all, for every 100 million miles driven in the US, there is only around one traffic death. Moreover, most are caused by reckless behavior. On the other hand, the data seems to suggest self-driving cars are involved in accidents more frequently than people, and that is with safer testing environments.
Hotz is not the only outspoken individual. A pioneer of the self-driving industry, Anthony Levandowski, also shared his opinions. He pointed out that “You’d be hard-pressed to find another industry that’s invested so many dollars in R&D and that has delivered so little.”
Levandowski went on to say, “We’re going to get there at some point. But we have such a long way to go.”
Maybe that is the truth. Perhaps we will get there, and perhaps it will be sooner than anyone thinks. But until then, it begs the question of how much more time and energy will go toward the pursuit of self-driving cars before anything comes of it.