Chapter 4
The Reinvention of the Automobile: From Mechanical Beast to Digital Marvel
In 1999, GM CEO Rick Wagoner posed a profound question: What would the car of the next hundred years look like? If automobiles were being invented today, what form would they take? This launched the most exciting research initiative of my career-reimagining mobility from first principles.
The result was "Autonomy," unveiled at the 2002 North American International Auto Show-a revolutionary concept vehicle that looked nothing like a conventional car. Just a sleek, six-inch thick skateboard-like chassis with four wheels, it housed a hydrogen fuel cell system with electric motors at each wheel. We demonstrated how different vehicle bodies could be easily swapped onto this universal platform.
The press loved it, with major publications calling it "the car of the future." We followed with the drivable Hy-wire prototype in 2003, which combined fuel cell technology with by-wire controls. My team made rapid progress on fuel cell technology, increasing power density sevenfold while reducing costs.
In 2005, a profound revelation occurred when we compared three disassembled vehicles: a Chevy Malibu with about 10,000 parts, a Toyota Prius hybrid with even more components, and our E-Flex Architecture prototype powered by a hydrogen fuel cell-which had only about 1,000 parts, one-tenth of the conventional vehicle.
When I brought Rick Wagoner to see this, he immediately grasped the implications: "This spells the end to the integrated auto industry as we know it." Electric vehicles would require far fewer employees to assemble, would eventually be cheaper to produce, and would open manufacturing to new competitors. Most critically, the expertise needed would shift from mechanical engineering to software development-a radical change for Detroit.
The internal combustion engine's days were numbered, though the transition would take longer than I initially predicted. As transportation shifts to service models with fleet operators owning vehicles throughout their 300,000-mile lifecycle, the economic case for electric propulsion becomes overwhelming.
Chapter 5
From Concept to Crisis: Detroit's Missed Opportunity
In October 2007, I had lunch with GM CEO Rick Wagoner, who was in high spirits despite the company's recent challenges. GM had just navigated difficult negotiations with the United Auto Workers regarding healthcare obligations. The deal freed GM from future healthcare obligations and cut starting wages for new employees in half. Our stock price jumped to a three-year high, and Rick was jubilant: "From here on out, I think it'll be much smoother sailing."
The irony was painful. Just as technology was finally making sustainable mobility possible-with advances in batteries, mapping, smartphones, sensing, and computing power-Detroit faced its biggest crisis ever from high gas prices and the subprime mortgage crisis.
As the auto industry's collapse accelerated through 2008, GM considered eliminating Saturn, Saab and Pontiac brands. The Big Three CEOs faced humiliation when they flew private jets to Washington to request bailout funds. President Bush arranged a $17.4 billion bridge loan for GM and Chrysler, but this merely postponed the inevitable.
In February 2009, industry sales hit a 28-year low, with GM sales dropping 53% compared to the previous year. On March 29, Rick Wagoner announced his resignation, with Fritz Henderson taking over as CEO. After thirteen years as my direct boss and close friend, Rick's departure devastated me. I immediately told my wife CeCe, "I've got to leave, too."
Just a week after Rick's resignation, we launched Project PUMA (Personal Urban Mobility and Accessibility) at the New York International Auto Show. This two-wheeled, self-balancing electric vehicle developed with Segway represented our vision for urban mobility-a 35-mile range on just 35 cents of electricity with a top speed of 35 mph. Despite its potential for congested urban environments, the press reception was brutal, with critics calling it "a rickshaw without charm" and "a pimped wheelchair."
By May 2009, the strain of GM's collapse had taken a physical toll on me. The Obama administration's restructuring plan included closing up to 20 factories and cutting 21,000 union jobs. I left GM on September 30, feeling branded with a scarlet "B" for bankruptcy after a lifetime with the company. My departure freed me to complete a book with Chris Borroni-Bird and MIT's Bill Mitchell called "Reinventing the Automobile: Personal Urban Mobility for the 21st Century."
Chapter 6
Google's Moonshot: The Secret Project That Changed Everything
After the DARPA Urban Challenge, autonomous vehicle development stalled despite the publicity and technological breakthrough. I bear responsibility for this missed opportunity-as GM's R&D chief, I declined Red Whittaker and Chris Urmson's proposal for a GM-Carnegie Mellon joint venture to continue the research, as GM was fighting for survival amid looming bankruptcy.
The next significant development came when Sebastian Thrun and Anthony Levandowski began working on Google's Ground Truth project to develop mapping technology. Levandowski built a self-driving Toyota Prius (dubbed "Pribot") for a Discovery Channel show that successfully navigated most of a route from San Francisco to Treasure Island autonomously-a remarkable achievement accomplished on a shoestring budget.
This demonstration coincided with Sebastian Thrun finishing his Ground Truth project at Google, where Larry Page persistently encouraged him to work on self-driving cars. Though initially resistant, claiming it couldn't be done, Thrun eventually couldn't provide Page with a technical reason why autonomous vehicles weren't feasible.
In October 2008, Thrun gathered an elite team of robotics experts at his Lake Tahoe chalet, including Mike Montemerlo, Dirk Haehnel, Hendrik Dahlkamp, Anthony Levandowski, Chris Urmson, and Bryan Salesky. Google's founders established two ambitious milestones: completing a series of ten challenging drives totaling 1,000 miles and accumulating 100,000 miles of autonomous driving on public roads.
Google's Street View project proved essential to the autonomous vehicle development. These detailed 3D maps became crucial for self-driving cars, allowing them to locate themselves precisely within inches by comparing real-time scans with stored landmark data.
The team meticulously programmed human-like driving patterns to make passengers feel comfortable. After confirming no laws explicitly prevented a computer from driving (as long as a human occupied the driver's seat), they conducted their first public road test in May 2009. Initial tests revealed the car was overcorrecting within its lane, but after adjustments, the vehicle drove more naturally, soon crossing milestones of 10, 100, and 1,000 autonomous miles.
The Chauffeur team taught their vehicle to recognize thousands of objects and behaviors through machine learning. By feeding the AI hundreds of thousands of images, the system learned to identify pedestrians, cyclists, children, animals, and even traffic cops directing traffic. Beyond mere perception, they developed a behavioral engine that could predict movements-anticipating that a cyclist might swerve into traffic to avoid a parked car or that a pickup truck would merge when lanes narrowed.
By summer's end of 2010, they'd completed nine of the ten Larry1K challenges when disaster threatened: a driver leaked information about the secret project to the New York Times. The team raced to complete the final challenge before the story broke, navigating winding mountain roads in fog, dodging debris, managing narrow roads, and completing the iconic descent down San Francisco's famous Lombard Street-just before the news broke.
When I experienced my first ride in Google's autonomous Prius in November 2010, I was nervous as I engaged the system on U.S. 101, one of America's busiest highways. When a Volkswagen cut us off, the Prius gently slowed to maintain a safe distance-it had anticipated the cut-off before it happened. Later, when passing a tractor-trailer, the Prius shifted slightly away in its lane, demonstrating sophisticated human-like behavior but with superior safety. I realized Thrun, Urmson, Levandowski, and Dolgov had created something that would change the world.
Chapter 7
The $4 Trillion Disruption: Reimagining Transportation Economics
During 2011, I investigated the economic impact of four converging disruptions: autonomous vehicles, purpose-designed vehicles, shared mobility, and electric propulsion. Working with the Earth Institute at Columbia University, I assembled a research team including mathematical modeling expert Bill Jordan.
Our initial analysis revealed Americans spend approximately $4.5 trillion annually on automobile transportation-$2 trillion in out-of-pocket costs ($0.65/mile) plus $2.5 trillion in time value ($0.85/mile), exceeding the federal government's budget.
We modeled an autonomous ride-sharing service in Ann Arbor, Michigan, requiring vehicles to arrive within two minutes of being summoned-matching the convenience of personal car ownership. Our mathematical models yielded astonishing results: the system could maintain two-minute response times with vehicles traveling empty just 5% of the time, achieving 75% utilization rates during operating hours, while requiring a fleet numbering only 15% of the population served.
Using Ann Arbor's real-world data of 200,000 personally owned vehicles making 740,000 daily trips, we calculated that just 18,000 shared autonomous vehicles could provide nearly instantaneous service for all internal trips. We verified these findings across multiple cities-consistently finding the 15% ratio held for communities with population densities above 750 people per square mile, covering most American urban areas.
Our cost analysis was equally revolutionary. A purpose-built two-person electric autonomous vehicle operating at scale would cost approximately $0.20 per mile-compared to the $1.50 per mile Americans typically paid for car ownership. This $1.30 per mile savings, multiplied across America's 3 trillion annual miles, represented a potential $3.9 trillion annual economic benefit.
When we presented these findings to Google's Chauffeur team in December 2011, the engineers were struck by the massive business opportunity-even capturing just 10% of total miles driven could generate $30 billion in annual profits. This presentation helped solidify Google's vision beyond just autonomous vehicles to a comprehensive mobility transformation combining shared fleets, electric propulsion, and purpose-built designs.
Chapter 8
The Great Awakening: How Silicon Valley Forced Detroit to Innovate
By late 2012, Google's Chauffeur team had developed a highway driver's assist product ready for testing. However, when they conducted internal trials, they discovered a critical flaw: the technology worked too well. Drivers completely disengaged from driving-working on laptops, applying makeup, and in one alarming case, falling asleep for 27 minutes while cruising at 60 mph.
This created a fundamental safety paradox. The system required drivers to remain alert to take control during unexpected situations, with only six seconds of warning. Yet the technology lulled users into a false sense of security that made them incapable of responding appropriately.
In December 2012, Chris Urmson gathered the Chauffeur team and announced they would abandon the highway assist approach. Instead, they would refocus on solving the complete autonomy problem-creating vehicles that could handle every aspect of driving without human intervention.
Urmson envisioned a vehicle specifically designed for autonomous transportation-as-a-service fleets. Unlike previous efforts that adapted existing vehicles, this would be a clean-sheet design explicitly built for autonomy. The resulting two-person pod, eventually named "Firefly," embodied the mobility disruption I'd envisioned.
The Firefly unveiling in May 2014 triggered an industry-wide stampede into transportation-as-a-service. After Google co-founder Sergey Brin's announcement, the New York Times published an article speculating that Google was developing driverless taxis.
Travis Kalanick, Uber's CEO, viewed autonomous vehicles as existential to Uber's future, since human drivers represented 70-90% of their cost per mile. At a conference, Kalanick declared: "When there's no other dude in the car, the cost of taking an Uber anywhere becomes cheaper than owning a vehicle."
The implications of autonomous technology were beginning to resonate throughout the industry. John Casesa, a respected auto industry investment banker, experienced his own epiphany about self-driving cars in 2014. After reading my Earth Institute paper projecting that just 9,000 autonomous taxis could provide superior mobility service in Manhattan at one-tenth the cost of human-driven cabs, Casesa was stunned. He described the coming change as "an earthquake for the industry" and emphasized: "This is not an evolution of your product. It's a substitution for your product."
By 2015, the industry finally started believing my long-standing prediction that self-driving technology would revolutionize automobiles. Uber's valuation surpassed General Motors, while Google became valuable enough to potentially buy GM outright with its cash reserves. This triggered widespread industry upheaval, with traditional automakers scrambling to catch up through acquisitions and partnerships.
GM made the most startling about-face after initially responding to Google with arrogance. CEO Mary Barra began listening to new voices, particularly president Dan Ammann, who observed, "Lots of people say they love to drive, but I haven't met anyone yet who says they love their commute." In early 2016, GM invested $500 million in Lyft, launched the ride-sharing company Maven, and spent $581 million to acquire forty-employee startup Cruise Automation for its autonomous expertise.
Chapter 9
The Human Factor: When Technology Outpaces Our Wisdom
On May 7, 2016, forty-year-old Joshua Brown became the first fatality involving autonomous vehicle operation when his Tesla Model S collided with a tractor-trailer while using Autopilot. Brown, a technology entrepreneur and former Navy bomb disposal expert, was an enthusiastic Tesla owner who had accumulated 45,000 miles in his "Tessy" and frequently posted videos demonstrating Autopilot's capabilities.
Multiple investigations revealed the Tesla didn't detect the tractor-trailer crossing its path. NHTSA's reconstruction suggested the trailer was visible for over ten seconds before impact, but Brown took no evasive action. Tesla offered two theories: either sensors mistook the white trailer for sky, or they detected it but classified it as a "false positive" like an overhead sign.
The author argues Tesla created dangerous conditions through several failures: naming the system "Autopilot," which created unrealistic expectations; inadequately informing users about the system's limitations; and Elon Musk personally retweeting one of Brown's videos, potentially endorsing his usage.
I took several lessons from Brown's death. Tesla and Musk have a pattern of overpromising and under-delivering, exemplified by Autopilot. I feel fortunate that Google's self-driving technology was developed under Page and Brin's funding before Musk entered the field. Google tested on public roads for seven years without a single crash implication, while Tesla's Autopilot was involved in a fatality just seven months after rollout.
The incident highlighted the fundamental challenge of partial automation-humans are terrible at monitoring systems that work well most of the time but occasionally require intervention. This is why Google pivoted to developing fully autonomous vehicles rather than driver assistance systems.
Chapter 10
The Road Ahead: Where Autonomy Takes Us Next
The early frustration felt by autonomous vehicle pioneers has given way to validation as their predictions have materialized. Red Whittaker, once battling university administration to develop his robots, is now celebrated on campus banners as a revolutionary. The initial dismissive attitudes between Silicon Valley engineers and Detroit automakers have transformed into collaborative partnerships.
Waymo's expansion strategy positions it not as a disruptor but as an enabler of future businesses through partnerships with companies who maintain their growing fleet. This approach helps Waymo scale at lower cost while delivering new mobility options.
I marvel at Larry Page and Sergey Brin's unwavering commitment to driverless vehicles-something traditional stakeholders in the 130-year-old transportation system would never have catalyzed with their vested interests in the status quo. The mobility revolution required visionaries with belief in digital technology, passion for compelling experiences, deep pockets, and commitment to improving the world.
To frame this as a race with winners and losers misunderstands the scale of what's happening. While technology companies with hundred-billion-dollar valuations may have advantages over automakers with mere tens of billions, the mobility disruption will ultimately benefit humanity broadly.
Children born today may learn to drive the way some still learn horseback riding-as a recreational skill rather than a necessity. The future won't be utopian, just as the internet brought trolls and fake news alongside its benefits, but cities will become more pleasant habitats with fewer inconveniences defining our daily routines.
If we succeed-and we will-we'll reduce annual traffic fatalities by 90 percent, eliminate oil dependence in transportation, free up urban land currently devoted to parking, reshape downtowns, democratize mobility access, and slow climate change. The future of mobility is coming-enjoy the ride!
![[5dfb54d8-9cb0-4b18-af3e-0862524f5a3c:c0001] By 2030, it'll be inconceivable to buy a car and not expe… p1-1](/_next/image?url=https%3A%2F%2Fd1y2du6z1jfm9e.cloudfront.net%2Fassets%2Fpodcast%2Fpurple.png&w=1024&q=50)
