Tesla’s former AI Director, Andrej Karpathy, recently made an appearance on Lex Fridman’s podcast. Karpathy is an AI researcher, engineer, educator and a founding member of OpenAI. He spoke with Fridman on a variety of topics, including AI, his departure from Tesla, and his interest in returning to the company.
Karpathy’s five-year tenure at Tesla was a pivotal part of the company's AI development. During his time with the company, he helped advance Tesla's self-driving system from lane-keeping on highways to partially autonomous driving on city streets.
After taking a four-month sabbatical, Karpathy parted ways with Tesla in July 2022. When Karpathy announced he was leaving Tesla, he said it was a difficult decision and thanked the Tesla Autopilot team, looking forward to seeing their momentum continue. Karpathy also stated that he wanted to spend more time on his long-term passions including AI technology, open source, and education.
From the podcast, we learned more details about what led to Karpathy's departure (1 hour, 44 minute mark). Apparently, his role at the company had developed into more of a managerial position, which drifted away from his passion for technical work in AI.
Karpathy said, “I think over time during those five years; I’ve kind of gotten myself into a little bit of a managerial position. Most of my days were, you know, meetings and growing the organization and making decisions about sort of high-level strategic decisions about the team and what it should be working on and so on.”
“And it’s kind of like a corporate executive role, and I can do it. I think I’m okay at it, but it’s not like fundamentally what I enjoy, and so I think when I joined, there was no computer vision team because Tesla was just going from the transition of using Mobileye, a third-party vendor, for all of its computer vision to having to build its computer vision system. So when I showed up, there were two people training deep neural networks.”
Elon Musk was quick to react to Karpathy's interview, stating on Twitter that "Andrej will always be welcome at Tesla."
Karpathy played a key role in Tesla's FSD development and it would be great to see him return for a second act. Based on his interview, it sounds like he may be interested in returning to Tesla to work on the Optimus, Tesla's humanoid robot.
Karpathy's Interview
You can watch Karpathy’s full interview with Lex Fridman below.
Subscribe
Subscribe to our newsletter to stay up to date on the latest Tesla news, upcoming features and software updates.
Tesla is getting ready to start rolling out FSD support for hardware 3 vehicles in China. Tesla is internally testing what appears to be a build of FSD V12.6 on HW3 vehicles with its employees.
A post on Chinese social media platform Rednote initially shared the software update details and was then shared by Tesla Newswire. We’ve verified that the vehicle in question is a 2023 Model Y, confirming it falls within the HW3 hardware generation.
We reviewed FSD V12.6 in depth here, and provided a comparison to FSD V13 here, but here’s the key points you’ll need to know.
FSD V12.6 is Tesla’s latest revision of FSD for HW3 vehicles and it’s is very comparable to FSD V13 in terms of features. It includes 3 speed profiles, Chill, Standard, and Hurry, which control how aggressive FSD is in making lane changes and ensuring it stays at, near, or above the speed limit.
This version of FSD also comes with an improved controller, as well as earlier and more natural lane change decisions, making it far more capable than V12.5 or V12.3 builds.
For Chinese customers getting their first taste of FSD after using Autopilot for years, it’s mindblowing. It can handle city level streets, hanlde turns and traffic control signs and get you to your destination with minimal interventions.
When Does It Release?
Given’s China’s more crowded streets and more complex road layouts, we weren’t sure whether Tesla would introduce support for HW3 vehicles in the market. This news is a positive and encouraging sign, not only for FSD in China, but for all HW3 vehicle owners out there.
Tesla usually releases updates to employees in advance of releasing them to its general customers - providing them with a second chance to find and fix bugs. Usually, once updates make their way to employee vehicles, and barring any major issues, they begin rolling out to customers within one to two weeks.
China currently requires owners to buy FSD, as FSD subscriptions aren’t available in the country yet. If you’re in China and bought FSD on a HW3 vehicle, it may not be long before you have access to FSD.
Tesla’s Cybertruck has officially earned a 5-Star Safety Rating from the NHTSA—an impressive achievement given the vehicle’s design. The achievement demonstrates Tesla’s engineering prowess. As one engineer points out, it wasn’t an easy feat.
Interestingly, the NHTSA only recently disclosed the results, despite the crash tests being completed a while ago. According to Lars Moravy, Tesla’s VP of Vehicle Engineering, the team had been aware of the 5-star rating for quite some time. While the reason for the delay remains unclear, now that the results are public, Tesla’s engineers can finally share how they achieved the rating.
Crumple Zones
Wes Morril, the Cybertruck’s Lead Engineer, wrote about the crash test video on X recently, addressing the claims that the Cybertruck doesn’t have a crumple zone. He also posted a side-by-side video (below) of the engineering analysis and the crash test itself.
Engineered Crash Safety
There’s a lot of engineering precision at play when a Cybertruck is involved in a crash. Unlike traditional crash structures that rely on crash cans and collapse points, the Cybertruck’s front gigacasting is designed to absorb and redirect impact forces in a highly controlled manner.
It all starts with the bumper beam, which crushes within the first few milliseconds of a high-speed impact. At the same time, the vehicle’s sensors rapidly analyze the crash dynamics and determine the optimal deployment of safety restraints, including airbags and seat belt pre-tensioners. These split-second actions are crucial in keeping occupants safe.
As the crash progresses, the vehicle’s structure deforms in a carefully engineered sequence. The drive unit cradle bends, directing the solid drive unit downward and out of the way, allowing the gigacasting to begin absorbing impact forces.
The casting crushes cell by cell, methodically dissipating energy in a controlled manner. This gradual deceleration reduces the g-forces transferred to occupants, making the crash much less severe. As the gigacast begins crushing, the safety restraints are deployed.
As Wes points out in his post - you can see how accurate the virtual analysis and modeling were. The video shows the simulated crash side by side with the real-life crash test and they’re almost identical. All that virtual testing helps provide feedback into the loop to design a better and safer system - one that is uniquely different than any other vehicle on the road.
All the armchair experts claimed the Cybertruck has no crumple zone and I get it, the proportions seem impossible. It was a tough one and there is a lot of engineering that went into it. Let me break it down for you: