CEO of Tesla, Elon Musk, recently tweeted that “excessive automation” at his company was a mistake that he is responsible for. He believes, “humans are underrated”, but fails to acknowledge that it is the auto industry doing all the under-rating. Musk’s own Tesla has been accused of under-reporting factory injuries and exploiting employees.
Musk has been a champion of greater automation, despite the crashes in January and March this year. Experts, on the other hand, have argued that automation is hard to achieve in the final assembly stage of production. Companies like Volkswagen and General Motors have tried, and more critically, failed.
Even, “the best producers – still the Japanese – try to limit automation”, according to analyst Max Warburton, who thought Tesla’s belief that they could automate most of the process of building Model 3 “a fallacy”.
The Tesla Problem
Musk’s inability to correctly predict the number of cars his company could produce should have been a sign of Tesla’s production problems. He slated the company to produce 100,000 models of S3 in 2017. At the end of the fourth quarter that year, Tesla had only made 2,425.
The NY Times and other news channels question the American car company’s future. But, Musk refuses to bow under pressure and continues to maintain that his company will be profitable in Q3 and Q4 this year.
More or Less Automation?
Tesla’s shortage in production has been attributed to less human workers in plants. Clearly, Musk and his company thought their machines more capable than they turned out to be.
After his tweet, however, Tesla is on a hiring spree. The plant in Fermont alone is hiring 126 people. But, will that be enough to solve Tesla’s problems?
The safety of the company’s cars is also in question. The problem, once again, seems to be the level of automation.
Tesla uses series automation in Autopilot. To be clear, it is not full automation. In series automation, technology only substitutes human drivers to perform repetitive tasks during the drive like maintain speed or lanes.
In case of more difficult tasks, however, human discretion is still necessary. The autopilot system may warn before giving control over, but it will not stop itself from doing so when it can no longer maintain control. The driver must be able to take control when this happens.
The problem with series automation is that it can sometimes be hard for drivers to stay focused. One report likened driving in Tesla’s Autopilot to staring at a clock all day. The task is boring to say the least. It also gets more difficult as the artificial intelligence gains experience. It becomes harder for drivers to preempt problems the more secure they feel using the technology.
The alternative is parallel automation. It is one that the aviation industry uses to improve the pilot’s flying experience. In this system, the driver would benefit from artificial intelligence. But, the help would be restricted. It would depend on the capabilities of existing self-driving systems.
Drivers would not only have more information about what to expect from their artificial intelligence, but also be able to use it better to enhance their driving experience.
Solving the Problem
The problem attributed to Tesla in this article is one of developing artificial intelligence. Series automation and parallel automation offer two types of systems. Both can lead us to fully-autonomous systems. Even if this stage is fairly in the distance. To gain full confidence in self-driving cars would require substantial experience – both for human and artificial intelligence – of “autonomous” vehicles. Until that day, we cannot underrate human or overrate artificial intelligence.