Debunking Self-Driving Myths: They’ll Be Here Tomorrow
Editor’s Note: In this series, Ground Truth asks industry experts to address some of the most pervasive misconceptions about self-driving technology. Here, Pedro Pacheco, Senior Research Director for Gartner’s CIO Research Group, tackles the “imminent arrival” myth.
For anybody listening to the hype about self-driving, it’s been the same song for years: Autonomous vehicles will be everywhere, soon.
In reality, many hurdles remain before self-driving technology is deployed at scale in cities around the world. Most realistic autonomous-vehicle developers know this, but don’t always admit it. But for pushers of the “autonomous vehicles are imminent” narrative, there are reasons for the hype: It keeps the public in suspense over a tantalizing technology; it spurs investors to open their wallets in expectation of outsized returns; and it compels industry observers to churn out think pieces and tweets that, in turn, keep the hype mill going and going.
These misunderstandings stem from many factors related to the novelty and complexity of the technology itself, but the main culprit is misguided public statements from some self-driving tech developers themselves. Though it will still be years before autonomous vehicles are ubiquitous on urban streets, that hasn’t stopped some companies from making pie-in-the-sky promises.
For Pedro Pacheco, a senior analyst at Gartner specializing in the autonomous vehicle industry, the misconceptions about the impending arrival of self-driving vehicles aren’t just adding to widespread confusion — they’re potentially undermining trust in the technology itself. “You have several companies making statements saying they plan a robotaxi service by this or that year,” says Pacheco, “and from the eye of the public, who are not AV experts, they think the service will start in that year.”
Of course, this kind of “we’re out in front” gamesmanship is nothing new, Pacheco says. “I remember when Mercedes-Benz back in the 1990s developed an autonomous vehicle, and they ran several trials in Germany.” The early display of autonomy elicited lots of breathless reactions among the media, analysts, and competitors, he says. But it took another 15-odd years for a Sebastian Thrun-led team to win the DARPA Grand Challenge and pave the way for the progress in self-driving that we’re seeing today.
Since that milestone moment in 2005, says Pacheco, “more and more players have come into the picture, the public hears more about the autonomous car, and this competitive tension arises, because all these players want to show that they are ahead of the game.” But this can be a bit of a catch-22. Self-driving companies often trade in proprietary technology, says Pacheco, which means they “can’t show all their cards.” Touting progress on one hand while maintaining secrecy on the other can create a morass of contradictory or confusing claims.
When these factors combine, it becomes hard for the public to follow what’s road-ready, what’s production-ready, what is a limited trial guided by teleoperation or safety drivers, and, critically, what’s actually safe for widespread use on public streets. “When it comes to full, production-level-ready, Level 4 cars,” he says, referring to the levels of autonomy established by SAE International, “the goalposts are not being set for how safety should be regulated.”
In the absence of consistent regulations, Pacheco goes on, self-driving vehicle developers are left to define what constitutes “safe” and when their products will cross that safety threshold. This leads companies to issue lots of claims that are hard for consumers to track or verify. “If you keep hearing claims in terms of timelines, and it doesn’t come, you might think, ‘Well they’ve been saying this for a while, and I still can’t use this technology, so it’s turning out to be harder than they imagine,’” Pacheco says. Such perceptions only hurt people’s trust in the technology’s safety, which is the linchpin to the successful deployment of self-driving vehicles.
Pacheco believes that autonomous vehicle developers should put an emphasis on realism and reachable benchmarks, like talking only about products when they are ready to go to market. “The market is starting to get a bit tired of claims of future milestones,” he said. “It has come to a point that making a claim saying it will happen in 2-3 years doesn’t mean much by itself, because there has been a lot of that happening and many claims that haven’t been fulfilled.”
There will come a time when we see “full production robotaxis around the world,” Pacheco believes. But more patience is required. “We’ll come to that this decade, but not this year,” he says.