Connect with us
tesla-ai-day-2022 tesla-ai-day-2022

News

Tesla AI Day News Roundup: Optimus, FSD Beta & Dojo updates

Published

on

Tesla AI Day has officially kicked off. Guests have started entering the venue already.

AI Day is an event mainly held to recruit talented people and welcome them to Tesla. However, it is still a Tesla event, so everyone expects some product surprises and updates, specifically about the company’s humanoid robot, Optimus, Dojo, and Full Self-Driving.

Teslarati will be closely following the event. This is our news roundup for 2022 AI Day, covering key information Tesla reveals at the event.

Photos and Videos aren’t allowed during the event from attendees. However, guests were able to capture some cool photos of a Tesla Semi with Cybertruck graffiti, a literal fork on the road, and some other cool set ups around the premises.

Tesla’s former AI Head, Andrej Karpathy, has brought out the (metaphorical) popcorn. His brief “comment” hints that AI Day 2022 might be as exciting as everyone anticipates.

Elon Musk set some expectations about Optimus, reminding everyone that during AI Day 2021, Tesla’s humanoid bot was just “a guy in a robot suit.” Musk also laid out the topics for AI Day 2022, saying that Tesla will talk about Autopilot and Dojo, too.

Advertisement

Optimus Takes the Stage

Tesla didn’t waste any time and brought out Optimus immediately. According to a Tesla mechanical engineer, AI Day 2022 is the first time Optimus has been “let out”– so to speak– without any external support.

“This is literally the first time the robot has walked on stage without a tether, on stage tonight,” Musk added. “The robot can actually do a lot more than we showed you. We just don’t want it to fall on its face.”

Tesla reveals videos of Optimus or in this case “Bumble-Cee” doing “work” around the Tesla office. Optimus carried a box from one area to another, watered plants, and even worked at the factory for a bit. The Tesla bot’s vision is very similar Autopilot.

Tesla also revealed Optimus’ potential final unit one production design. “Our goal is to make a useful humanoid robot as soon as possible,” said Elon Musk. The Tesla CEO also shared that Tesla aims to make Optimus’ price less than $20,000 or cheaper than a car.

Tesla is using some of the technology in its car in Optimus’ body as well, such as the battery pack, cooling system and more. The company also uses that same technology it uses for its cars to simulate Optimus’ movements and reactions to external collisions.

Tesla is basing Optimus’ body design on the human body. The company has been closely studying the structure of the human body while making the humanoid robot’s overall design. For instance, Tesla designed Optimus’ hands with the idea that factories worldwide are designed ergonomically, or optimized for the human hand. Teslarati briefly covered the significance of robots’ hands in a previous article, linked below.

FSD Beta Updates

The Tesla FSD Beta now has 160,000 customers, compared to 2,000 customers in 2021. Tesla is expected to release v.10.69.2.3 after AI Day, although a precise roll out date has not be announced yet.

Tesla explained the progress its made with Full Self-Driving Beta. The Tesla FSD experts explained how the Full Self-Driving makes decision to AI day guests and the role that customer data played to refine the software. The company also explained occupancy and the role it plays with 3D mapping and providing a birds-eye-view for the car. Tesla is working hard to optimize its video model training as well.

Tesla also talked a lot about its FSD Lane Networks during AI Day 2022. In the past few weeks, Teslarati has received reports from FSD testers, who specifically brought up issues with lane selection. To see “under the hood”–so to speak–somewhat explains the lane issues FSD testers experience on the road.

After multiple test loops and drives, there’s really just one main problem remaining for me at this point on 10.69.2, it’s significant, and that is lane selection,” noted long-time FSD tester Les. 

Advertisement

Tesla has developed a new auto-labeling machine to help with 3D labeling. The FSD experts explained how the software uses other clips to fill out the picture under certain conditions when the camera shows an unclear picture.

Tesla also talked a bit about simulation. The experts showed how it could simulate worlds or environments, using the data gathered from its fleet. It revealed a simulation of San Francisco that was created within two weeks by one employee. Tesla may update a simulated world quickly and as updated develop.

Dojo Updates

Tesla’s goal with Dojo is to build a single accelerator. A key step to realizing its goals was its training tile, which it unveiled during AI Day 2021. Tesla has been trying to figure out how to make its Dojo design scalable and has run into challenges along the way. However, the company’s “fail fast” mindset has helped it push through road blocks and move forward.

The Dojo team showed images of a Cybertruck and Semi running on Mars using stable diffusion achieved through Dojo.

Tesla experts explained that Dojo reduced work that would usually take months to a single week.

Tesla plans to build its first Exapod by 2023, which is expected to significantly increase its autolabeling output . It will be the first Exapod of 7 that Tesla plans to build in Palo Alto.

Tesla ended AI Day 2022 by answering questions from attendees. Tesla hopes that through their thorough explanations during the event, the company would be seen as more than an automaker. And, of course, Tesla hopes that its AI Day 2022 presentation also entices talented individuals to join the company.

The Teslarati team would appreciate hearing from you. If you have any tips, contact me at maria@teslarati.com or via Twitter @Writer_01001101

Maria--aka "M"-- is an experienced writer and book editor. She's written about several topics including health, tech, and politics. As a book editor, she's worked with authors who write Sci-Fi, Romance, and Dark Fantasy. M loves hearing from TESLARATI readers. If you have any tips or article ideas, contact her at maria@teslarati.com or via X, @Writer_01001101.

Advertisement
Comments

News

Tesla FSD’s new Mad Max mode is getting rave reviews from users

It does appear that Mad Max mode is destined to be one of the system’s biggest steps forward to date.

Published

on

Credit: Whole Mars Catalog/X

Tesla’s release notes for the newly released Mad Max mode for FSD (Supervised) V14.1.2 simply stated that the feature “comes with higher speeds and more frequent lane changes than Hurry.” But as per videos that have been posted online by FSD users who have tested the system, it does appear that Mad Max mode is destined to be one of the system’s biggest steps forward to date. 

It is then no surprise that the new capability is getting rave reviews from Tesla owners. 

Impressive tests

A look at posts on social media platform X would show that, similar to past FSD releases, numerous Tesla content creators immediately tested Mad Max mode on real-world streets after it was downloaded onto their vehicle. Considering that the update was released rather late, the first tests of Mad Max mode were done at night. Despite this, it was evident that Tesla worked very hard to make Mad Max mode into something that is very useful in real-world scenarios.

This could be seen in videos from longtime Tesla owner @BLKMDL3, who observed that Max Max mode was “amazing” and like “perfect for LA traffic” due to its cautious but assertive nature. Later on, the Tesla owner noted that after eight drives, it was evident that FSD (Supervised) V14.1.2 was impressive. 

Assertive but safe

Other testers such as Model Y owner Sawyer Merritt noted that Mad Max mode drives very quickly and confidently, with smoother acceleration that is still very safe. These were echoed by another longtime FSD tester, Dirty Tesla, who noted that Mad Max mode seems to be designed for heavy, aggressive traffic so users could fit in better. The FSD user did, however, observe that Mad Max mode does speed up a lot on open roads. 

Advertisement

Recent comments from Tesla AI Head Ashok Elluswamy have indicated that Mad Max mode was created to be a solution for daytime congested traffic, which is arguably one of the most soul-crushing experiences that drivers deal with on a daily basis. With this in mind, it does appear that FSD (Supervised) V14.1.2 could prove to be a notable step forward in Tesla’s push towards true autonomous driving.

Continue Reading

Elon Musk

Elon Musk highlights the biggest flaw in X’s monetization program

Elon Musk also stated that YouTube manages creator payments “much better.”

Published

on

MINISTÉRIO DAS COMUNICAÇÕES, CC BY 2.0 , via Wikimedia Commons

Elon Musk has admitted that X’s creator payout system isn’t living up to expectations, and he has highlighted the current system’s biggest flaw. 

Amidst complaints about low and inconsistent payments, the platform’s owner acknowledged that X has been “underpaying and not allocating payment accurately enough.” Musk also stated that YouTube manages creator payments “much better.”

Musk acknowledges payout issues

Recent discussions about the social media platform’s payout issues began when X product head Nikita Bier stated that the company was developing new upgrades for “power users.” This prompted X user Peter Duan to raise ongoing concerns about being “consistently underpaid” compared to his peers. Bier responded candidly, suggesting that “creator payouts do more harm than good and we need to off-ramp to a different system.”

Musk then weighed in on the matter, contradicting Bier’s view. “No,” Musk wrote in his reply, “the issue is that we are underpaying and not allocating payment accurately enough. YouTube does a much better job.” The Tesla CEO’s comment immediately reignited debates about X’s monetization program, which some have criticized for its rather unpredictable nature.

X’s monetization challenges

Since X launched its ad revenue-sharing program in 2023, the system has promised to reward Premium subscribers who generate high engagement with verified accounts, as noted in a WION report. Creators, however, have argued that the company’s payout model has remained inconsistent, with revenue fluctuating even when view counts stay stable. Reports have noted that some users with millions of monthly impressions have received just a few hundred dollars.

Advertisement

By contrast, YouTube’s Partner Program, which takes a 45% cut of ad revenue, is known for more transparent and predictable payments. Musk’s admission that YouTube handles monetization more effectively could then hint at a potential shift towards a new monetization program for X, a platform that has become increasingly critical to social conversations over the years. 

Continue Reading

News

Tesla exec hints at FSD Mad Max mode’s killer feature

The release notes of Tesla’s v14.1.2 FSD update indicate that Mad Max mode “comes with higher speeds and more frequent lane changes than Hurry.”

Published

on

Credit: @BLKMDL3/X

Tesla may have just rolled out its boldest Full Self-Driving (FSD) upgrade yet, but the company’s Head of AI, Ashok Elluswamy, hinted at the recently released “Mad Max” mode’s actual killer feature.

As per the Tesla executive, FSD’s Mad Max mode is designed to provide drivers with optimum driving performance during what are commonly the most tedious driving conditions on real-world roads.

Where Mad Max mode truly shines

Tesla drivers and longtime FSD users responded positively to the rollout of Mad Max mode. The performance of the update was so notable that @WholeMarsBlog, a longtime FSD tester, described it as epic. The FSD tester’s comments were posted on X as videos of Mad Max mode’s real-world performance were being shared online.

In response to the Tesla owner and longtime FSD tester, Elluswamy noted that drivers would probably love Mad Max mode even more during daytime hours, when traffic is denser. “You’ll love it more during day time / denser traffic. Really showcases its decision making,” the Tesla executive wrote in his post.

The release notes of Tesla’s v14.1.2 FSD update indicate that Mad Max mode “comes with higher speeds and more frequent lane changes than Hurry.” Videos shared online showed that Mad Max mode, despite its assertive driving style, is still a very cautious and safe driver, similar to past FSD releases.

Advertisement

Made for real-world traffic and long commutes

Traffic congestion typically peaks during daytime hours, when drivers could at times spend hours navigating crowded intersections and fast-changing lanes. For many Tesla owners, having an FSD mode that can confidently manage that chaos could be a game-changer.

Simply put, the feature’s extra assertiveness could allow Mad Max mode to excel in the kind of traffic that tests even the most patient drivers. By improving decision-making in those conditions, the company may be positioning FSD as a true solution for the everyday stress of stop-and-go commutes, packed freeways, and unpredictable city driving.

The “Mad Max” name itself isn’t new. Elon Musk first teased it back in 2018 as a playful nod to aggressive freeway driving. Its reappearance in Tesla’s modern FSD system, however, hints at the notable maturation of Tesla’s autonomous driving efforts over the years.

Continue Reading

Trending