Tesla AI Day: what to expect from Elon Musk’s latest big announcement

It’s been almost two years since Tesla’s first Autonomy Day event, in which CEO Elon Musk made many lofty predictions about the future of autonomous vehicles, including his infamous claim that the company has “a million robots on the road” by the end. of 2020. And now it’s time for the second part.

This time the event will be called “AI Day” and according to Musk, the “only goal” is to persuade experts in the field of robotics and artificial intelligence to come and work at Tesla. The company is known for its high turnover, the most recent being Jérôme Guillen, a key executive who worked at Tesla for 10 years before resigning recently. Attracting and retaining talent, especially big names, has proven to be a challenge for the company.

The August 19 event is scheduled to begin at 5:00 p.m. PT / 8:00 p.m. ET at Tesla’s headquarters in Palo Alto, Calif. According to an invitation obtained by Electrek, it will include “an Elon talk, hardware and software demonstrations from Tesla engineers, Model S Plaid tests, and more.” Much like Battery Day, the event will be streamed live on Tesla’s website, giving investors and media, as well as the company’s many fans, a glimpse of what is in development.

Musk and other senior company officials are expected to provide updates on the rollout of Tesla’s Full Self-Driving (FSD) Beta 9, which began reaching more customers this summer. We can also get details on Tesla’s “Dojo” supercomputer, the training of its neural network, and the production of its FSD computer chips. And there will also be “a glimpse of the future of AI at Tesla beyond our fleet of vehicles,” the invitation reads.

Let’s start with what we know and work our way up to speculation about what’s to come.

FSD deployment

The big news from Tesla’s first Autonomy Day was the introduction of the company’s first computer chip, a 260-square-millimeter piece of silicon that Musk described as “the best chip in the world.” Originally, Musk claimed that Tesla’s cars wouldn’t need any hardware updates, just software, on the road to full autonomy. Turns out that wasn’t exactly the case; they would need this new chip – two of them, in fact – to eventually drive themselves.

A lot has happened between the 2019 event and now. Last month, Tesla started shipping live software updates for FSD beta v9, its long-awaited driver assistance system, certainly not standalone, but certainly advanced. This means that Tesla owners who bought the FSD option (which now costs $ 10,000) could finally use many of Autopilot’s advanced driving assistance features on local non-road streets, including Navigate on Autopilot, Auto Lane Change, AutoPark, Invocation, and traffic light and stop control.

The update won’t make Tesla’s cars fully autonomous, nor will it launch “a million self-driving cars” on the road, as Musk predicted. Tesla owners who have fully autonomous driving should always pay attention to the road and keep their hands on the wheel. Some do not, which can have tragic consequences.

Loved by fans, hated by security advocates, FSD software recently put Tesla in hot water. In recently posted emails between Tesla and the California Department of Motor Vehicles, the company’s Autopilot software director made it clear that Musk’s comments (including his tweets) do not reflect the reality of Tesla’s vehicles. can actually do. And now Autopilot is under investigation by federal regulators who want to know why Teslas with Autopilot continue to crash into emergency vehicles.

In addition to rolling out the beta of FSD v9, Tesla has also had to adapt to the global chip shortage. On a recent earnings conference call, Musk said the company’s engineers had to rewrite some of their software in order to support other computer chips. He also said that Tesla’s future growth will depend on a swift resolution of the global semiconductor shortage.

Tesla relies on chips to power everything from its airbags to modules that control vehicle seat belts. It is not clear whether the FSD chips, which are produced by Samsung, are affected by the shortage. Musk and his cohort could give a glimpse of that at this week’s event.

Dojo

Outside of the car, Tesla uses a powerful supercomputer to train the AI ​​software which is then delivered to its customers via live software updates. In 2019, Musk teased this “super powerful training computer”, which he called “Dojo”.

“Tesla is developing a [neural net] training computer called Dojo to process very large amounts of video data ”, he then tweeted. “It’s a beast!

He also hinted at Dojo’s computational power, claiming he was capable of an exaFLOP, or quintillion (1,018) of floating-point operations per second. It is an incredible power. “To match what an exaFLOP IT system can do in a second”, NetworkWorld wrote last year, “you would have to perform a calculation every second for 31,688,765,000 years.”

For comparison, chipmaker AMD and computer maker Cray are currently working with the US Department of Energy on the design of the world’s fastest supercomputer, with a processing power of 1.5 exaFLOP. Dubbed Frontier, AMD claims the supercomputer will have as much processing power as the next 160 fastest supercomputers combined.

When completed, Dojo should be among the most powerful supercomputers on the planet. But rather than performing advanced calculations in areas like nuclear and climate research, Tesla’s supercomputer is running a neural network in an effort to train its AI software to power self-driving cars. Ultimately, Musk said Tesla would make Dojo available to other companies who want to use it to train their neural networks.

Earlier this year, Tesla Chief AI Officer Andrej Karpathy gave a presentation at the 2021 Computer Vision and Pattern Recognition Conference, where he gave more details about Dojo and its network. neurons.

“For us, computer vision is the bread and butter of what we do and what enables autopilot,” Karpathy said, according to Electrek. “And for this to work really well, we have to master fleet data, train massive neural networks, and experiment a lot. So we have invested a lot in the calculation.

Other robots?

Earlier this month, Dennis Hong, founder of UCLA’s Robotics and Mechanisms Lab, tweeted a photo of a computer chip that many speculate is the internal hardware used by the Tesla Dojo.

But Hong is also an interesting character for other reasons. He specializes in humanoid robots and took part in the DARPA Urban Challenge which launched the race for autonomous cars. (His team placed third.)

Asked on Twitter whether his lab worked with Tesla, Hong posted funny emojis but declined to comment. We could learn more about how Hong’s work and Tesla’s business intersect on AI Day.

Musk has expressed his desire for Tesla to become more than just an auto company. “I think in the long run people will think of Tesla as much as an AI robotics company as we are an automotive company or an energy company,” he said earlier this year.

The future

A warning to anyone tapping into the AI ​​Day livestream: Take Musk’s predictions of near-term accomplishments with a huge grain of salt. The items that will be discussed at this event are unlikely to have a measurable impact on the business of the company in the months to come.

Self-driving cars are an incredibly difficult challenge. Even companies like Waymo, which are perceived to have the best autonomous vehicle technology, are still struggling to get it right. Tesla is no different.

“A key question for investors will be what is the latest timeline for achieving full autonomy,” said Gene Munster, managing partner of Loup Funds, in a note. “Despite Elon’s ambitious target at the end of this year, our best guess is that 2025 will be the first year of public availability of Level 4 autonomy.”

The rest of 2021 is already busy for Tesla. The company is to open factories in Texas and Germany. And he must equip the production of his highly anticipated Cybertruck, which has been delayed until 2022. Full autonomy, as it is, can wait.