Here’s What I Learned at LA’s Autonomous Vehicle Hype Conference

My day at an autonomous vehicle conference in Los Angeles started in a human-driven Uber.

On a Wednesday morning, Waymo said it would take at least 15 minutes to snag a car for a five-minute ride to NeueHouse Hollywood, where Ride AI was hosting its first-ever autonomous vehicle summit, featuring some of the sector’s leading players.

I wasn’t particularly in a time crunch, but, like most people, I don’t call a rideshare driver half an hour or more before I leave the house to get to a destination.

Exit app. Open Uber.

It’s fair to say that, at least for now, the battle between human and robot drivers is still raging.

Stephen Hayes, Lyft’s VP of autonomous operations, and Ryan Green, CEO of Gridwise — a platform for rideshare drivers to track their earnings — both said at the conference that the future of ride-hailing will look like a hybrid market of human and robot drivers.

Green estimated that that’ll be the case for at least the next 10 to 15 years.

It may be Hayes’ and Green’s prerogative to make such a claim. But that outlook made perfect sense to me — even as someone who has had great experiences in a Waymo — and appeared to resonate throughout the summit.

A hybrid future

Toyota Research Institute CEO Gill Pratt jumpstarted the event as the first speaker to say it’s a “myth that we need autonomy to help with terrible human driving.”

“Humans are actually reasonably safe, pretty good drivers. For every hundred million miles there’s an accident,” Avinash Balachandran, TRI’s VP of the Human Interacting Driving division, told me. “Where we really see value for autonomy is this ability to help drivers in situations where they tend to struggle.”

For a conference centered entirely on the future of autonomous mobility, I was surprised that some of the top players in the AV field weren’t throwing out rosy pictures or estimates of when robots will take over the wheel and our city roads.

Tesla, whose CEO has made wild predictions like one million robotaxis by 2020, was noticeably absent from the conference but still came up during panel talks.

Instead, voices from various autonomous mobility players, from ride-hailing to OEM software providers, spoke on the challenges of perfecting full self-driving technology and the still-fresh business opportunities for the human-driven vehicle segment.

Amnon Shashua, CEO of Mobileye, talked about the lack of “precision” in AI, stemming partly from the need to perfect the application of redundancies and a lack of data.

Over at Wayve, a UK-based autonomous driving software company, Kaity Fischer, the VP of commercial and operations, talked about the “very large untapped market” of level two and level three assisted and autonomous driving systems for consumer vehicles.

Both levels still need humans behind the wheel but with varying degrees of driver input.

“For the foreseeable future, there’ll be a very long period of time where it’s a hybrid solution of vehicle ownership intermingled with varying levels of autonomy, and those things all need to be able to coexist,” Fisher told me when I asked her what she thought city streets will look like decades down the line. “So levels two, three, and four all need to be able to share a space on the road and operate safely in conjunction.”

Christoph Lütge, an economist and AI ethics expert at the Technical University of Munich, told me that in Germany, where robotaxis exist only for limited testing, Level 3 driving is still not widely implemented.

“It should have happened already, but it’s not really happening,” he said.

Riding the hype cycle

This is not to underestimate the progress folks in San Francisco and other markets have experienced firsthand. One of the reasons everyone gathered in Los Angeles for an autonomous vehicle conference is undoubtedly because of Waymo.

As Sophia Tung, a former engineer and now an AV-focused content creator, told me, we’ve seen an autonomous vehicle hype cycle before — this time’s just a bit different.

Timothy B. Lee, one of the Ride AI moderators who has reported on transportation for Ars Technica before launching his Substack, Understanding AI, pointed to the Gartner hype cycle: Technological innovation first comes with the skyrocketing of what’s called peak “inflated expectations” before crashing into a “trough of disillusionment,” then rising again, more gradually, on a “slope of enlightenment” followed by a “plateau of productivity.”

The peak of inflated expectations happened around 2016 or ’17, Lee told me, before an Uber autonomous test vehicle killed a woman in Tempe, Arizona, in 2018.

“Then there was a period of three or four years, from 2018 to 2022 or so, when things were going very badly for these companies,” he said. “You had Uber shut down their program, Lyft shut down their program. Ford had Argo (AI) that shut down. That was a bad time to be in the industry and not as much fun to cover. And then, the last two years I would say it’s been on the upswing pretty much since Waymo started expanding to San Francisco.”

Lee’s prognosis is that the great human-driver replacement will be gradual. In the next 20-plus years, city roads will see fewer human drivers before they completely disappear.

For example, he cited the decades it took before automatic elevators made human elevator operators obsolete.

“There’s going to be a long process,” he said.

My day in Los Angeles ended with another human-driven rideshare, which arrived in the same amount of time it took to say goodbye to my friends, put on my shoes, and walk out of their apartment.

The closest Waymo was 13 minutes away.

Exit app. Open Uber.