Tesla AI chief decides to take a 4-month vacation, netizens are worried

Tesla’s AI chief, Musk’s humerus, Andrej Karpathy, was reported to have submitted a rare vacation, a four-month long vacation.

Netizens expressed concern, because before this,Tesla senior engineer Doug Field never came back after taking more than a month off.If Karpas is gone, Tesla’s Autopilot and FSD will undoubtedly be greatly affected.

May be worried about this, Musk simply announced the news on Twitter,And to clarify the truth – Karpas really just took a vacation.

Karpas himself also said that this period is just for good rest and relaxation, looking forward to returning to work quickly.

Who is Andre Capas?

Andrei Karpas was born in Czechoslovakia in 1986 and moved to Toronto, Canada with his family at the age of 15.

Karpas has a very rich experience in studying. He has been studying in two different countries and three different universities. In 2009, he graduated from the University of Toronto with a bachelor’s degree in computer science and physics, a master’s degree in computer science from the University of British Columbia in 2011, and a doctorate degree in computer science from Stanford University in 2016.

After graduating from a master’s degree, Karpas interned at Google for a period of time, mainly engaged in unsupervised deep learning, supervised deep learning, computer vision and other related work.

During his Ph.D. at Stanford University, Karpas was under the tutelage of Professor Feifei Li of the Stanford Vision and Learning Laboratory. Later, Karpas also served as a lecturer for the university’s deep learning course, which was offered at Stanford for the first time in history, and was designed and taught by him personally. Neural Networks”. Later, the number of people attending the class increased from 150 in 2015 to 750 in 2017.

Approaching his Ph.D. graduation, in January 2016, Karpas joined OpenAI, which was initiated by Musk, and became one of the founding members of the company. At first, he was mainly responsible for recruiting employees, and later served as a research scientist in the company, responsible for generating models. deep learning and deep reinforcement learning.

You know, OpenAI is a non-profit artificial intelligence research organization whose purpose is to promote and develop artificial intelligence for the benefit of mankind.

At the beginning of its establishment, it was to benchmark DeepMind, and later it did produce sensational AI achievements such as GPT-3.

After joining Tesla

After working at OpenAI for a year and a half, Karpas joined Tesla, and until now,As Tesla’s AI lead, deep learning and computer vision expert, leading Tesla Autopilot’s computer vision team.

AI vision, at Tesla, is critical to FSD’s fully autonomous driving capabilities. This set of vision system can process the video data collected by 8 cameras to make a high-definition three-dimensional map, which can be used to reflect the driving speed and direction of surrounding people and things.

So what exactly does the team do?

Focus on data collection and neural network training, etc. First, data collection needs to be done, such as internal data labels, custom label interfaces, etc.; the second is to train neural networks; the third is to find ways to make it work, such as segmentation, detection, 3D/depth prediction, etc.; fourth It is to arrange the production and deployment of chips.

“Today, Autopilot makes driving safer and more convenient, but our team’s goal is to accelerate the development of FSD capabilities and deploy them to millions of our own vehicles,” said Capas.

In short, Karpas is like Musk’s left arm, influencing Tesla’s autonomous driving and the future and direction of various AI products.

From the current point of view, Tesla’s Autopilot is not doing well enough. The U.S. Highway Traffic Administration has identified at least 12 major traffic accidents related to Autopilot.

And last Saturday, the Tesla FSD Beta launched in Canada, marking the expansion of FSD functionality beyond the United States.

It is worth mentioning that Musk also announced that it plans to launch its “Optimus Prime” humanoid robot Tesla Bot in 2023, which is next year.

So why taking a vacation is so concerned, mainly because of the time point and duration, which is a bit thought-provoking.

Why take time off during a critical period?

Karpas himself gave the reasons. He said that the reason for taking the leave is mainly for two purposes:One is to take a good rest and relax, and travel by the way; the other is to spend some time on self-improvement.

On Twitter, he wrote: “I have been working at Tesla for nearly five years, and now I want to take a break and take a good rest, travel, etc. During this time, I can focus on self-improvement, such as strengthening myself. Let’s train a neural network. Although I miss Tesla’s robots and the Dojo supercomputing cluster, and I can’t wait to see them!”

Regarding Karpas’ latest update, he tweeted: “I’m now on a ‘digital nomad’ trip, pack my bags and head east, saying hi to friends along the way, reading Paper, knocking Typing code or something. Still in the UK, then to Europe, Asia, and finally back to the San Francisco Bay Area.”

In short, “I’ll go back to Tesla.”

Goals after returning from vacation?

After returning from a four-month sabbatical, Karpas has two important projects to complete.

One is the deployment of the Dojo supercomputer, which will be operational next year.

This computer will set a world record, replacing Japan’s “Fudake” as the world’s fastest supercomputer.

The supercomputer can store massive amounts of video data, which it then uses to train neural networks for autonomous driving, all to speed up Autopilot and FSD.

The other is to update the hardware of FSD and launch the 4.0 version, the computing power can reach 4 times that of the original 3.0 version, and it can be installed on Tesla’s Cybertruck in the future.

Version 2.5 was launched in 2017, and version 3.0 was launched in 2019. It has been 3 years since the last version update.

When it comes to this 4.0 version, Musk said he hopes it can drive safely, even better than humans. But it is still improving at this stage, and I hope to develop more powerful computers in the future.