Many companies are pursuing self-driving cars, but only one is on a deadline to make it work before it runs out of cash: Uber.
The ride-sharing company was the first to deploy self-driving cars on pubic roads (Pittsburgh, Pennsylvania, in late 2016), and is now the first to record a fatal accident. Depending on what investigators discover about that accident that struck and killed a homeless woman in Tempe, Arizona, on March 18, the company could set back development of the technology in the US.
Uber is cooperating with authorities, and the mayor’s office released a statement after the incident declaring, “public safety is our top priority.” But examining the events leading up to this week’s accident suggests that the company may have rushed in too fast, even as companies such as Tesla, Waymo, Ford and GM are close behind.
True to form, Uber cut corners before the Tempe accident. In its drive to get Uber to customers, it defied local officials, betting (correctly) that it could lobby for regulatory change afterwards. It was the same pattern it followed in Pittsburgh, where its relationship soured as Uber’s demands mounted and benefits to the public didn’t and in California, where the state yanked the company’s vehicle registrations after Uber ignored demands it register its automated vehicles before launching its 16-car pilot in San Francisco.
Like officials before him, Arizona’s governor Doug Ducey initially welcomed the startup, signing an executive order permitting testing, and Uber promptly shipped its vehicles west to patrol the state’s roads.
Prior to the fatal crash, Uber CEO Dara Khosrowshahi predicted that within 18 months the company would send out fully autonomous Ubers to pick up some riders in Phoenix (all of them have safety drivers behind the wheel today). Within five years, it would have perfected self-driving transportation in Arizona’s biggest city, and then rolled the technology out globally a decade later.
As of February, Uber had logged about 50,000 rides with self-driving technology on board. The company has also apparently opted to have only one safety driver in the car, rather than the two people (one engineer to monitor the cars’ sensors and systems) now being used by most other companies.
Uber remains under pressure to deliver on its technology. In 2016, the company is thought to have burned through $3 billion (despite earning profits from its top cities). Founder and former CEO Travis Kalanick has called self-driving cars “basically existential” for Uber—the technology may reduce fares by 80%.
Khosrowshahi has tried to put the company on the straight and narrow, but financial realities as well as Uber’s risk-taking culture are pushing the company to roll out the technology as fast as possible. Meanwhile the competition is right behind.
General Motors’ Cruise Automation will launch a car without a steering wheel next year. Waymo’s autonomous vehicles in the Phoenix metropolitan area (which includes Tempe) are picking up passengers. Ford told Quartz this week it had no plans to change its autonomous trials. Tesla, despite enduring a fatal accident involving Tesla’s Autopilot in the US, has upgraded its safety functions rather than scale back its technology (driver error was found to be partially at fault). Only Uber and its planned partner, Toyota, appear to have suspended their testing programs on public roads.
Uber’s dicey reputation makes it likely that the entire industry could slam on the brakes with only a few more bad incidents, argues Bryant Walker Smith, a law professor at the University of South Carolina, who studies the implications of autonomous vehicles. “This incident was uncomfortably soon in the history of automated driving,” writes Smith. “In the United States, there’s about one fatality for every 100 million vehicle miles traveled, and automated vehicles are nowhere close to reaching this many real-world miles. This arguably first fatality may not tell us much statistically, but neither is it reassuring.”
Uber needs to show regulators that its standards are unimpeachable. “[The company] shouldn’t even touch its onboard and offboard systems unless credible observers are present,” Smith writes. “In this crash, a multitude of data will likely be available to help understand what happened—but only if those data can be believed.” Officials across the country are re-evaluating their testing programs and both Boston and California have issued statements.
For now, Americans see self-driving cars as inevitable within the next ten to 50 years, but least than half of them would actually want to ride in one. Why? Lack of trust remains the top concern.
Cars kill 103 people daily on US roads, according to The National Highway Traffic Safety Administration, and more than 94% of crashes are due to driver error (pdf). While autonomous vehicles promise to be safer, cheaper, and more convenient, people are far less likely to tolerate errors by machines than humans. We will need a new social contract to let them into our lives, write researchers in Nature Human Behavior on the psychology of self-driving vehicles. “This social contract will be bound as much by psychological realities as by technological and legal ones,” they write.