There will come a time as you drive down the road, perhaps in the very car you now own, when a car without a driver or anyone inside will pass in the opposite direction. That transition period — certainly 10 years, more likely 20 and perhaps 30 — when normal cars and driverless cars share the roads figures to be the most revolutionary automotive era since Karl Benz put the first production car on the road in 1885.
"This is going to be a really dangerous time," said Paul Brubaker, head of ATI21, a consortium of transportation technology innovators. "When I talk to the folks that I know who are developing the artificial intelligence that's going to enable this, that's the biggest challenge that they've got, predicting the unpredictable behavior of the human drivers."
The era when drivers and driverless cars will share the road looms larger this year as both houses of Congress ponder regulations to govern autonomous vehicles and as fleets of test vehicles are mixing it up with hands-on-the-wheel drivers in several states.
It's easier to predict and prepare for how autonomous cars will interact with each other than it is to imagine every move — some of them demonstrably idiotic — that a human driver might make.
That's where the soft-spoken, even-tempered Iris may come in.
Iris is not her real name, but she is a real person (a.k.a. Heather Caruso), and she took part in a remarkable psychological study about how people might react to cars without steering wheels and pedals.
Give the car a human name and a human voice and people who ride in it are more likely to relax and trust it.
This is not a radical concept. Humanizing inanimate objects — B.B. King named his guitar Lucille and probably forgave her for a bad note — is not new, but neither has it factored much in the monumental transition from driving ourselves to being driven by a machine with a computer.
That may well be because the people developing autonomous cars are automotive engineers, not psychologists. They are laser-focused on creating cars that — as nearly as they are able — won't make mistakes.
It's not their worry that the number of people who say they would be happy to ride in a fully autonomous car dropped by 11 points to just 13 percent this year, according to an MIT survey. Another survey found that almost half of people said they would "never purchase a car that completely drives itself."
In other words, they won't trust it.
"We're trying to figure out, what are those human kind of interactions that engender trust between a human and a machine?" said Jack Weast, Intel's chief architect for self-driving solutions. "What we are finding is the more you can embody those kind of human trust interactions into a human-machine experience, the human perception of the safety and security of that machine changes dramatically."
Iris came to life in a study published in the Journal of Experimental Social Psychology by Nicholas Epley, Joy Heafner and Adam Waytz. (As befits a journal devoted to psychology, their report used the word "anthropomorphize" rather than "humanize.")
"When people attributed a mind to a [car], they tended to think of it as more intelligent, more thoughtful, more capable of doing the things it's intended to do," Epley said. "If you think that it can think — it's not just a mindless machine, but you kind of get a sense that it's a thoughtful person — then you're more likely to trust it."
Epley, who teaches behavioral science at the University of Chicago Booth School of Business, and his colleagues gave Iris a genuine human voice and, as a result, a gender. Then they used driving simulators to test whether 100 people reacted differently when dealing with Iris and a nameless, voiceless but otherwise identical autonomous car.
They found that riders were roughly four times as comfortable with Iris than with the other vehicle.
"The presence of a voice is really critical for inferring the presence of a mind," Epley said. "When you hear what somebody has to say, people are judged to be more mindful, more thoughtful, more rational."
Then they tossed a curveball at their test group. The car they were riding in got into a crash it couldn't avoid. It was sideswiped by another vehicle that clearly was at fault. People riding in the non-humanized autonomous car were more than four times as likely to blame the computer in their car than the people in Iris were.
"We found that people were more relaxed physiologically [with Iris] when they got into an accident," Epley said. "People who thought their car was intelligent, thoughtful and rational blamed their car less because they assumed it was more that other car's fault."
With a public already skeptical about the wisdom of fully autonomous cars, reaction to the initial mishaps they get into figures to play a huge role in determining how quickly Americans get comfortable with the new cars.
There are two approaches in play. Traditional automakers, who need to sell cars each year, plan to gradually introduce features until the day arrives when they've produced a fully autonomous vehicle. Newcomers to automaking, such as Waymo (whose auto development wing once was called Google), plan to put fully autonomous vehicles on the road from Day One.
There also are different views within the industry on whether the first cars will debut on everyday streets and highways or in more limited circumstances.
"Where you have cars driving with no steering wheel, no pedals, I think first what you're going to see is designated areas of cities that are going to be cordoned off," Weast said. "Think about 40 city blocks — let's say the entire downtown [of Washington], D.C. — off-limits to human-driven cars, only automated robo-taxis cruising around."
Fully automated cars work best in the right conditions, so they might pop up first in places where the weather is best suited to their capabilities.
"Right now the technology to operate in all weather, all conditions isn't there," ATI21's Brubaker said. "We've got some development to do. We've got some testing to do."
Said Weast: "You always have to qualify it, because if you truly mean any road, anywhere in the world under any weather condition, under any scenario, we're probably quite a ways off from that. Maybe we're a decade or more from that kind of thing."
Blizzard-prone areas won't be the first to get autonomous cars, Weast said, but in other places they will arrive "much, much sooner than people expect."
"We definitely will see deployments of highly and fully automated vehicles within the next five years, for sure," he said.
Karl Brauer, executive publisher of Cox Automotive, likens the arrival of autonomous cars to that of the iPhone.
"The next 10 years of personal transportation will be like the last 10 years of mobile technology," Brauer said. "Ten years ago the iPhone still hadn't been introduced. That's pretty amazing when you think about it. That sort of paradigm shift is what autonomous tech will do to our society in the next 10 years."
Epley still worries.
"The most important thing that's going on in this transitional period is that people tend not to be forgiving when computers make mistakes," he said. "It's totally stupid in the sense that these cars are actually much safer on the road."