Tesla’s chief executive officer and technoking, Elon Musk, is not a man associated with pessimism or defeatism. Yet, after years in which he boldly declared that Tesla would develop a self-driving car, Musk has finally admitted what industry insiders had long known: developing a safe, reliable self-driving car is a hard problem. The Verge discusses Musk’s admission as well as the predictable fawning response from the massive Musk fanbase.
To anyone who has spent a modicum of time thinking about self-driving cars, these wondrous things are not easy to develop and are not likely to achieve commercialisation within the near to medium term. Yet Musk has consistently pushed the idea that Tesla is on the verge of developing a safe, reliable self-driving car. Despite peddling a notion that serious thinkers knew was rubbish, the Twitteratti responded to Musk’s admission with the usual fawning, back-patting, and encouragement. Tesla has successfully created a narrative that it’s cars are fully self-driving. Indeed, it has an “autopilot and full self-driving” feature which is a misnomer, to say the least, because a fully self-driving car does not need any input from the driver. If you believe your Tesla truly has “full self-driving” capability, well, good luck to you.
Musk is the king of spin. He regularly oversells and under delivers in many aspects of Tesla’s business. This has made Tesla’s stock highly volatile and led some experts to believe that the underlying economics of the business do not justify its lofty valuation. The “Autopilot and Full Self-Driving” feature is just another example of how Musk prioritises spin over substance. In 2018, he said that version 9 would appear by August. It didn’t. In 2019, he declared that Tesla would have more than a million full self-driving cars on the roads. Well, we know that hasn’t happened. Recently, he said that FSD 9 beta would be shipped out soon, adding, “I swear!” to assuage the doubters.
Musk gets more credit than he deserves. The difficulties of self-driving cars have been known for a long time. Developing a fully-self-driving car -not what Tesla is selling as full self-driving- requires solving a host of artificial intelligence problems. Tesla’s admission that realizing Musk’s dream is harder than they realised is just an admission of a reality that pretty much anyone who follows the industry has understood for a long time. Consider that Google’s Waymo project, which many experts see as being one of the frontrunners in developing self-driving cars, has not made the same ambitious promises as Musk.
Tesla appears to be ahead of the pack primarily because, not only does it make bold proclamations, unlike its competitors, it is willing to test beta versions of its technology on its customers. The argument is that by testing beta versions on customers, it can generate a lot of real-world data that it can use to develop the system. Tesla’s customers are enthusiastic about being the company’s guinea pigs. Many of them request to be white listed to test beta versions of the “Autopilot and Full Self-Driving” feature. This has given Tesla the air of being a company ahead of its time. Even if other firms are ahead of them in terms of real development.
Tesla regularly reports that its “Autopilot and Full Self-Driving” feature is safe, though this is based on curated data and self-selected metrics, rather than objective tests. This is a bit like grading yourself in an exam rather than going out and getting actual, objective Botox certification. Importantly, for anyone who reads the fine print or has actually experienced a Tesla, the feature works only with the active participation of the driver. Despite this reputation for innovation, Waymo is way ahead of Tesla, having actual self-driving cars on public roads. A crucial difference between Waymo and other self-driving cars and Tesla, is that Waymo and Tesla’s other competitors, use a combination of sensors -lidar, radar, cameras and others- to provide a robust framework to guide the car. Tesla only uses a camera. It’s no wonder they are behind.