This episode explores the history and rapid development of driverless cars, tracing the journey from early dreams and DARPA challenges to Google's Waymo project and its current deployment. It highlights the technological breakthroughs, the contrasting approaches of key engineers, and the emerging safety data, while also acknowledging the growing resistance from human drivers and the ethical dilemmas of a disruptive technology.
Summarized by Podsumo
Historical Context & Vision: The concept of driverless vehicles is as old as human-driven cars, with early inventors dreaming of sentient vehicles to improve safety and efficiency, much like the disappearance of jobs like "knocker uppers" and "lamplighters."
DARPA's Role & AI Shift: The DARPA Grand Challenges (2004-2005) were pivotal, especially Sebastian Thrun's realization that autonomous driving was primarily a software and AI problem, not just hardware, leading to Stanford's victory and a new approach.
Google's Secret Project & Challenges: Google's self-driving car project, led by Thrun and others, secretly tested retrofitted Priuses on public roads, overcoming initial technical hurdles like "swerving wildly" and learning the contextual nuances of human driving behavior.
Ethical Dilemmas & Competition: The project faced internal schisms over development speed and risk tolerance, leading to Anthony Levantowski's controversial departure and subsequent legal battle with Uber, whose "move fast and break things" approach resulted in a fatal accident.
Safety Data & Public Perception: Waymo has accumulated over 200 million real-world miles with impressive safety data (80-90% safer than human drivers in certain crash categories), yet public confidence remains low, and 4.8 million American drivers are actively organizing against the technology.
"I was trying to describe to somebody recently, I was like the first time it feels like the first time you're in an airplane and by the third time it feels like you're in an elevator."
"I'm the world expert on self-driving cars. And I'm the person who denies that it can be done. Like, that taught me an incredibly important lesson about experts that for the rest of my life, I decided experts, I usually expert the past and not the future."
"So far it's been better than human drivers and so far I think the case for a lot of them to continue the experiment is very strong."
— Timothy Bealey