Science

Americans Agree on Self-Driving Cars With One Big Exception

There's one big exception.

Tesla

It’s likely a matter of time before self-driving cars rule the road, with the first fully autonomous vehicles potentially hitting the market before 2020. The advent of cars that drive themselves will mean a major rethink of everything we think we know about driving, from our laws to our cultural expectations of what it even means to be inside a car. It won’t be an easy adjustment. Here’s the good news: According to new research shared with Inverse, Americans are more ready than you might expect for the age of driverless cars — with one or two big exceptions.

Australia’s National Transport Commission got the ball rolling on these issues in a big way earlier this month when it released this report suggesting — with a few caveats — that the occupants of self-driving cars should be exempt from DUI laws.

To better understand the average American outlook on these questions, the company Unanimous A.I. created a hive mind. Instead of a traditional poll — one of which last month found Americans are still uneasy around self-driving cars — Unanimous A.I. instead uses what’s known as swarm intelligence to bring people together and have them make a group decision on a topic. Let’s take a look at an example from this research, in which the participants were asked: “An occupant in a fully automated self-driving car is best described as _?”

Unanimous A.I.

The hive mind — which comprised seven Republicans, nine Democrats, and 15 independents — overwhelmingly saw the occupant of an autonomous car to be a rider a lot more than a driver. And you’re seeing a replay of what the swarm participants saw live. The idea is that as individuals within the swarm see how other people are thinking, this gives them a psychological push toward what becomes the group decision. Total unanimity isn’t always possible — the collective brainpower behind this decision was 80 percent of the hive mind, which isn’t actually super high — but the swarm in this case effectively simulates how society at large might evolve in its thinking.

“When it comes to the change in cultural norms from drastic technological change, it’s always a challenge to find solutions that are universally accepted across a large population,” Unanimous A.I. founder Louis Rosenberg tells Inverse. “That’s why artificial swarm intelligence is so useful in this application, as it converges on the sentiment that resonates most strongly across the population as a whole.”

Looking at some of the swarm’s decisions, it’s possible to chart out a pretty clear consensus on where people are headed in terms of thinking about self-driving cars. There was one major area of indecision that we’ll get to in a moment. But first, let’s look at where the hive mind agreed:

Unanimous A.I.

The swarm overwhelmingly felt autonomous cars would make the roads safer, with the majority believing such vehicles would make them a lot safer.

Drunk Drivers

Unanimous A.I.

On the question the Australian commission had taken up, the hive mind wasn’t fully convinced occupants of autonomous cars should be exempt from DUI laws, as the swarm ultimately only agreed to this with low confidence. But as the animation shows, again most of the disagreement was between those who said “no” with high or low confidence. The answer itself wasn’t especially in doubt.

Unanimous A.I.

Either way, the swarm had little doubt that driverless cars would reduce DUI-related accidents. Other tests revealed the swarm believed intoxicated occupants of an autonomous car were far more like people giving directions to a cabbie than someone actually operating a vehicle, and the hive mind gave only a 10 percent chance that an impaired person could manage to alter the vehicle’s operation.

That’s where the swarm agreed. Now let’s consider where it really was of many minds.

“I think the most contentious swarm I saw in this particular topic this week was whether occupants of a fully automated self-driving car should have liability for accidents that such a vehicle might get into,” Unanimous A.I. researcher Chris Hornbostel tells Inverse. “Eventually it was decided that the rider wasn’t responsible. However, later in the swarm, the group thought that the entity who needed to carry insurance on a fully automated self-driving car was the owner, and not the vehicle manufacturer.”

Unanimous A.I.

As Hornbostel indicates, the swarm does eventually say riders are never liable, but the collective brainpower behind this decision is shockingly low, just 45 percent. It’s not that very many, if any of the swarm participants were arguing riders are liable, but those who believed exceptions should exist remained steadfast in that belief, even as the swarm pulled toward the “never liable” option. These might be nuances, but for the participants of the swarm, these nuances in terms of who is and isn’t responsible are absolutely critical.

“In those cases, I think we caught a glimpse of how the public will struggle with this topic and putting away old mores about liability and who bears responsibility,” says Hornbostel. “The swarm was able to look to the future pretty easily when imagining that riders in a fully-automated car shouldn’t bear liability… but then still held to current insurance responsibility regulations in thinking that owners of a vehicle whose driving instructions are fully programmed by a manufacturer should still assume full insurance burdens on the car.”

The driverless future figures to be strange and unpredictable, as our entire concept of what transportation is sees its most dramatic transformation in at least a century. Good to know that we’re mostly ready to face that tomorrow — give or take the question of what to do with our insurance cards.

If you liked this article, check out this video about spherical tires for self-driving cars.

Related Tags