Elon Musk introduced Ashok Elswami, an Indian-born engineer who was Tesla's first hire on the self-driving team and now leads social media platform X (formerly Twitter). In his letter of appreciation, the tech billionaire praised the automaker's “success in AI” and his work on self-driving software.
What Elon Musk said about Ashok Elswami
Musk wrote: “Thanks, Ashok!”
Ashok was the first to join the AI/Autopilot team at Tesla and eventually went on to lead all of the AI/Autopilot software.
Without him and our amazing team, we would be just another automotive company looking for an autonomous driving supplier that doesn't exist.
FYI, I never told him to say anything and I didn't even know he wrote this until I saw it 10 minutes ago!”
In the post, Musk also shared a link to an article written by Elswamy of X, which highlighted how the Tesla CEO has been a “leading advocate for AI and autonomy at Tesla.”
What Ashok Elswami said about Elon Musk and AI
Elluswamy wrote: “he [Musk] They have always pushed us to achieve great things, even if the ideas seemed impossible at the time.”
Back in 2014, Autopilot started on a ridiculously tiny computer with about 384KB of memory and poor computing (it didn't even have native floating point arithmetic). He asked his engineering team to implement lane keeping, lane changing, longitudinal vehicle control, curvature, etc. Many people on the team thought the request was crazy. But he never gave up and pushed his team to achieve this very challenging goal. Against all odds, Tesla shipped the world's first Autopilot system in 2015. It would be many years before a close second came to market.
In 2016, Tesla began doing all of the computer vision needed for Autopilot in-house, rather than relying on outside vendors. Many thought it was foolhardy to bet our product on developing a vision system from scratch in a few months when it had taken other companies over a decade to do it. However, we accomplished this goal in less than 11 months. This was a strategically important move and began the development of a strong AI team at Tesla.
He not only promoted powerful AI software, but also powerful AI hardware. Tesla, thought of by others as just a car company, manufactured custom silicon to efficiently run neural networks. This hardware was originally designed in 2017 and went into production in February 2019, making it highly competitive compared to hardware announced to date. For reference, this 5-year-old AI computer has roughly 8x the AI inference computing of the most advanced Apple M3 chip. It is still capable of running modern end-to-end neural networks built on the latest AI techniques.
He is the one who solved the problem of autonomy by betting on vision and AI, rather than relying on the crutches of sensors and high-resolution maps. It may be obvious to anyone who has experienced the latest version of FSD that you can see everything that matters and drive the car based on pure vision. But before 2020, it wasn't obvious to most people. In fact, many “experts” in the field mocked Tesla and Elon for these choices. We proved them wrong by shipping supervised FSD to millions of cars, and showed that with good AI software, a car can handle the complexities of city driving, such as turning, handling intersections, and yielding to pedestrians, just by looking outside. In fact, we got rid of radar and ultrasonic to really focus on the core of the problem: AI. It is almost paradoxical that today, Tesla has the fewest raw sensors compared to any production car, yet the most autonomous capabilities. It was only because of his strong conviction and deep understanding of the problem that we were able to make such a reverse bet.
He began work on the humanoid robot at Tesla in 2021, before ChatGPT and other clear examples of the rise of AI. Like vehicle autonomy, Optimus is being developed to be capable, scalable and cost-effective for broad global contributions.”
What Elon Musk said about Ashok Elswami
Musk wrote: “Thanks, Ashok!”
Ashok was the first to join the AI/Autopilot team at Tesla and eventually went on to lead all of the AI/Autopilot software.
Without him and our amazing team, we would be just another automotive company looking for an autonomous driving supplier that doesn't exist.
FYI, I never told him to say anything and I didn't even know he wrote this until I saw it 10 minutes ago!”
In the post, Musk also shared a link to an article written by Elswamy of X, which highlighted how the Tesla CEO has been a “leading advocate for AI and autonomy at Tesla.”
What Ashok Elswami said about Elon Musk and AI
Elluswamy wrote: “he [Musk] They have always pushed us to achieve great things, even if the ideas seemed impossible at the time.”
Back in 2014, Autopilot started on a ridiculously tiny computer with about 384KB of memory and poor computing (it didn't even have native floating point arithmetic). He asked his engineering team to implement lane keeping, lane changing, longitudinal vehicle control, curvature, etc. Many people on the team thought the request was crazy. But he never gave up and pushed his team to achieve this very challenging goal. Against all odds, Tesla shipped the world's first Autopilot system in 2015. It would be many years before a close second came to market.
In 2016, Tesla began doing all of the computer vision needed for Autopilot in-house, rather than relying on outside vendors. Many thought it was foolhardy to bet our product on developing a vision system from scratch in a few months when it had taken other companies over a decade to do it. However, we accomplished this goal in less than 11 months. This was a strategically important move and began the development of a strong AI team at Tesla.
He not only promoted powerful AI software, but also powerful AI hardware. Tesla, thought of by others as just a car company, manufactured custom silicon to efficiently run neural networks. This hardware was originally designed in 2017 and went into production in February 2019, making it highly competitive compared to hardware announced to date. For reference, this 5-year-old AI computer has roughly 8x the AI inference computing of the most advanced Apple M3 chip. It is still capable of running modern end-to-end neural networks built on the latest AI techniques.
He is the one who solved the problem of autonomy by betting on vision and AI, rather than relying on the crutches of sensors and high-resolution maps. It may be obvious to anyone who has experienced the latest version of FSD that you can see everything that matters and drive the car based on pure vision. But before 2020, it wasn't obvious to most people. In fact, many “experts” in the field mocked Tesla and Elon for these choices. We proved them wrong by shipping supervised FSD to millions of cars, and showed that with good AI software, a car can handle the complexities of city driving, such as turning, handling intersections, and yielding to pedestrians, just by looking outside. In fact, we got rid of radar and ultrasonic to really focus on the core of the problem: AI. It is almost paradoxical that today, Tesla has the fewest raw sensors compared to any production car, yet the most autonomous capabilities. It was only because of his strong conviction and deep understanding of the problem that we were able to make such a reverse bet.
He began work on the humanoid robot at Tesla in 2021, before ChatGPT and other clear examples of the rise of AI. Like vehicle autonomy, Optimus is being developed to be capable, scalable and cost-effective for broad global contributions.”
