Artificial Intelligence Is Here To Calm Your Road Rage

I am behind the wheel of a Nissan Leaf, circling a parking lot, trying not to let the day’s nagging worries and checklists distract me to the point of imperiling pedestrians. Like all drivers, I am unwittingly communicating my stress to this vehicle in countless subtle ways: the strength of my grip on the steering wheel, the slight expansion of my back against the seat as I breathe, the things I mutter to myself as I pilot around cars and distracted pedestrians checking their phones in the parking lot.

“Hello, Corinne,” a calm voice says from the audio system. “What’s stressing you out right now?”

The conversation that ensues offers a window into the ways in which artificial intelligence could transform our experience behind the wheel: not by driving the car for us, but by taking better care of us as we drive.

Before coronavirus drastically altered our routines, three-quarters of U.S. workers—some 118 million people—commuted to the office alone in a car. From 2009 to 2019, Americans added an average of two minutes to their commute each way, according to U.S. Census data. That negligible daily average is driven by a sharp increase in the number of people making “super commutes” of 90 minutes or more each way, a population that increased 32% from 2005 to 2017. The long-term impact of COVID-19 on commuting isn’t clear, but former transit riders who opt to drive instead of crowding into buses or subway cars may well make up for car commuters who skip at least some of their daily drives and work from home instead.

Longer commutes are associated with increased physical health risks like high blood pressure, obesity, stroke and sleep disorders. A 2017 research project at the University of the West of England found that every extra minute of the survey respondents’ commutes correlated with lower job and leisure time satisfaction. Adding 20 minutes to a commute, researchers found, has the same depressing effect on job satisfaction as a 19% pay cut.

Switching modes of transit can offer some relief: people who walk, bike or take trains to work tend to be happier commuters than those who drive (and, as a University of Amsterdam study recently found, they tend to miss their commute more during lockdown). But reliable public transit is not universally available, nor are decent jobs always close to affordable housing.

Technology has long promised that an imminent solution is right around the corner: self-driving cars. In the near future, tech companies claim, humans won’t drive so much as be ferried about by fully autonomous cars that will navigate safely and efficiently to their destinations, leaving the people inside free to sleep, work or relax as easily as if they were on their own couch. A commute might be a lot less stressful if you could nap the whole way there, or get lost in a book or Netflix series without having to worry about exits or collisions.

Google executives went on the record claiming self-driving cars would be widely available within five years in 2012; they said the same thing again in 2015. Elon Musk throws out ship dates for fully autonomous Teslas as often as doomsday cult leaders reschedule the end of the world. Yet these forecasted utopias have still not arrived.

The majority of carmakers have walked back their most ambitious estimates. It will likely be decades before such cars are a reality for even a majority of drivers. In the meantime, the car commute remains a big, unpleasant, unhacked chunk of time in millions of Americans’ daily lives.

A smaller and less heralded group of researchers is working on how cars can make us happier while we drive them. It may be decades before artificial intelligence can completely take over piloting our vehicles. In the short run, however, it may be able to make us happier—and healthier—pilots.

Lane changes, left turns, four-way stops and the like are governed by rules, but also rely on drivers’ making on-the-spot judgments with potentially deadly consequences. These are also the moments where driver stress spikes.

Many smart car features currently on the market give drivers data that assist with these decisions, like sensors that alert them when cars are in their blind spots or their vehicle is drifting out of its lane.

Another thing that causes drivers stress is uncertainty. One 2015 study found commuters who drove themselves to work were more stressed by the journey than were transit riders or other commuters, largely because of the inconsistency that accidents, roadwork and other traffic snarls caused in their schedules. But even if we can’t control the variables that affect a commute, we’re calmer if we can at least anticipate them—hence the popularity of real-time arrival screens at subway and bus stops.

The Beaverton, Ore.-based company Traffic Technology Services (TTS) makes a product called the Personal Signal Assistant, a platform that enables cars to communicate with traffic signals in areas where that data is publicly available. TTS’s first client, Audi, used the system to build a tool that counts down the remaining seconds of a red light (visually, on the dashboard) when a car is stopped at one, and suggests speed modifications as the car approaches a green light. The tool was designed to keep traffic flowing—no more honking at distracted drivers who don’t notice the light has turned green. But users also reported a marked decrease in stress. At the moment, the technology works in 26 North American metropolitan areas and two cities in Europe.

TTS has 60 full- and part-time employees in the U.S. and Germany, and recently partnered with Lamborghini, Bentley and a handful of corporate clients. Yet CEO Thomas Bauer says it can be hard to interest investors in technologies that focus on improving human drivers’ experience instead of just rendering them obsolete. “We certainly don’t draw the same excitement with investors as [companies focused on] autonomous driving,” Bauer says. “What we do is not quite as exciting because it doesn’t take the driver out of the picture just yet.”

Pablo Paredes, a clinical assistant professor of psychiatry and behavioral sciences at the Stanford School of Medicine, is the director of the school’s Pervasive Wellbeing Technology Lab. Situated in a corner of a cavernous Palo Alto, Calif., office building that used to be the headquarters of the defunct health-technology company Theranos, the lab looks for ways to rejigger the habits and objects people use in their everyday lives to improve mental and physical health. Team members don’t have to look far for reminders of what happens when grandiose promises aren’t backed up by data: Theranos’ circular logo is still inlaid in brass in the building’s marble-floored atrium.

It can be hard to tell the lab’s experiments from its standard-issue office furniture. To overcome the inertia that often leads users of adjustable-height desks to sit more often than stand, one of the workstations in the team’s cluster of cubicles has been outfitted with a sensor and mechanical nodule that make it rise and lower at preset intervals, smoothly enough that a cup of coffee won’t spill. In early trials, users particularly absorbed in their work just kept typing as the desk rose up and slowly stood along with it.

But the millions of hours consumed in the U.S. each day by the daily drive to work hold special fascination for Paredes. He’s drawn to the challenge of transforming a part of the day generally thought of as detrimental to health into something therapeutic. “The commute for me is the big elephant in the room,” he says. “There are very simple things that we’re overlooking in normal life that can be greatly improved and really repurposed to help a lot of people.”

In a 2018 study, Paredes and his colleagues found that it’s possible to infer a driver’s muscle tension—a proxy for stress—from the movement of their hands on a car’s steering wheel. They’re now experimenting with cameras that detect neck tension by noting the subtle changes in the angle of a driver’s head as it bobs with the car’s movements.

The flagship of the team’s mindful-commuting project is the silver-colored Nissan Leaf in their parking lot. The factory-standard electric vehicle has been tricked out with a suite of technologies designed to work together to decrease a driver’s stress.

On a test drive earlier this year, a chatbot speaking through the car’s audio system offered me the option of engaging in a guided breathing exercise. When I verbally agreed, the driver’s seatback began vibrating at intervals, while the voice instructed me to breathe along with its rhythm.

The lab published the results of a small study earlier this year showing that the seat-guided exercise reduced driver stress and breathing rates without impairing performance. They are now experimenting with a second vibrating system to see if lower-frequency vibrations could be used to slow breathing rates (and therefore stress) without any conscious effort on the driver’s part.

The goal, eventually, is a mass-market car that can detect an elevation in a driver’s stress level, via seat and steering wheel sensors or the neck-tension cameras. It would then automatically engage the calming-breath exercise, or talk through a problem or tell a joke to ease tension, using scripts developed with the input of cognitive behavioral therapists.

These technologies have value even as cars’ autonomous capabilities advance, Paredes says. Even if a car is fully self-driving, the human inside will still often be a captive audience of one, encased in a private space with private worries and fears.

Smarter technologies alone aren’t the solution to commuters’ problems. The auto industry has a long history of raising drivers’ tolerance for long commutes by making cars more comfortable and attractive places to be—all the while promising a better driving experience that’s just around the corner, says Peter Norton, an associate professor of science, technology, and society at the University of Virginia and author of Fighting Traffic: The Dawn of the Motor Age in the American City. From his perspective, stress-busting seats would join radios and air conditioners as distractions from bigger discussions about planning, transit and growing inequality, all of which could offer much more value to commuters than a nicer car.

In addition, how long it will be before these latest features become widely available options is an open question. Paredes’ lab had to suspend work during the pandemic, as it’s hard to maintain social distancing while working inside of a compact sedan. TTS is in talks to expand its offerings to other automakers, and Paredes has filed patents on some of his lab’s inventions. But just because a technology is relatively easy to integrate in a car doesn’t mean it will be standard soon. The first commercially available backup cameras came on the market in 1991. Despite their effectiveness in reducing collisions, only 24% of cars on the road had them by 2016, according to the Insurance Institute for Highway Safety, and most were newer luxury vehicles. (The cameras are now required by law in all new vehicles.)

These technologies also raise new questions of inequality and exploitation. It’s one thing for a commuter to opt for a seat that calms them down after a tough day. But if you drive for a living, should the company that owns your vehicle have the right to insist that you use a seat cover that elevates your breath rate and keeps you alert at the wheel? Who owns the health data your car collects, and who gets to access it? All of the unanswered questions that self-driving technologies raise apply to self-soothing technologies as well.

Back in Palo Alto, the pandemic still weeks away, I am piloting the Leaf around the parking lot with a member of the lab gamely along for the ride in the back. The chatbot asks again what’s stressing me out. I have a deadline, I say, for a magazine article about cars and artificial intelligence.

The bot asks if this problem is “significantly” affecting my life (not really), if I’ve encountered something similar before (yep), if previous strategies could be adapted to this scenario (they can) and when I’ll be able to enter a plan to tackle this problem in my calendar (later, when I’m not driving). I do feel a little better. I talk to myself alone in the car all the time. It’s kind of nice to have the car talk back.

“Great. I’m glad you can do something about it. By breaking down a problem into tiny steps, we can often string together a solution,” the car says. “Sound good?”