autoevolution
 

Why the FSD Beta Software Was Put On the Back Burner at the Tesla AI Day

Tesla’s AI Day was not about the Optimus robot or its Bumble C cousin, despite the two humanoid machines taking the front stage and the headlines. Elon Musk made it clear that this was a recruitment event. However, Tesla FSD Beta software was an equally important part of it. Surprisingly, Elon Musk did not use the occasion to claim that Tesla FSD would be ready “next year” as he usually did.
Why the FSD Beta software was on the back burner at the Tesla AI Day 11 photos
Photo: Tesla
Why the FSD Beta software was on the back burner at the Tesla AI DayWhy the FSD Beta software was on the back burner at the Tesla AI DayWhy the FSD Beta software was on the back burner at the Tesla AI DayWhy the FSD Beta software was on the back burner at the Tesla AI DayWhy the FSD Beta software was on the back burner at the Tesla AI DayWhy the FSD Beta software was on the back burner at the Tesla AI DayWhy the FSD Beta software was on the back burner at the Tesla AI DayWhy the FSD Beta software was on the back burner at the Tesla AI DayWhy the FSD Beta software was on the back burner at the Tesla AI DayWhy the FSD Beta software was on the back burner at the Tesla AI Day
Tesla’s event in Palo Alto, California, was more of an inside story, with the people invited being carefully vetted. It was, mainly, a recruitment event, as Elon Musk said before, an open invitation to tech kings to join the Tesla AI team. Or, as Musk put it, “to convince some of the most talented people in the world, like you, guys, to join Tesla and help make it a reality.”

But, besides that, it was a chance to look at what is brewing inside Tesla’s labs. Not just the Bumble C and the Optimus Bot, which Musk said would be ready 5-10 years from now (we all know what this means), but also the FSD Beta system, which was supposed to be production-ready years ago. Remarkably, neither Elon Musk nor other AI team members ventured to give a deadline for when the Full Self-Driving would be fit to drive a car without a human driver behind the steering wheel.

This is new, considering how Musk has promised driverless cars every year since at least 2014. Not having a deadline must be, indeed, liberating, and hopefully, something good will come out of that. But don’t be fooled. This doesn’t mean that Tesla or Elon Musk has abandoned plans to put self-driving cars on the road. They are, in fact, closer than ever, as the progress with the FSD Beta 10.69.2.2 showed.

On the sidelines of AI Day, Musk’s team touched the FSD advances and what they mean for both Tesla vehicles and the future robots the company plan to sell in the next 5 to 10 years. Reportedly, Tesla’s most critical advance is the revolutionary automated labeling system. While the name doesn’t say much, this is about identifying objects encountered by Tesla vehicles, including other cars, pedestrians, road signs, and whatnot.

In the past, the Tesla Autopilot team would watch tons of videos to identify and describe objects in the clips captured by cameras and sensors on Tesla vehicles. Road boundaries, lane markings, and pedestrians obstructing the full view of a stop sign would then be identified and labeled accordingly.

This labeling work is instrumental in training Tesla’s neural network computers to enable Tesla cars to navigate traffic and avoid obstacles. Now, Tesla uses an automatic-labeling technology, which allows it to digest half a million video clips each day. This significantly boosts the Autopilot/FSD team, with countless situations analyzed in less time than before. And yet, on this AI Day, Elon Musk refrained from making any big promises regarding the automated driving capabilities of Tesla’s vehicles.

This is intriguing, especially as Tesla’s advance toward self-driving seems tangible after many years of overpromising and underdelivering. Plenty of reports show that the latest iteration of the Tesla FSD software appears to work as advertised most of the time. With the auto labeling software in place and the advancement in machine learning, Elon Musk would’ve been entitled to claim that FSD software would be soon ready for deployment, probably next year.

Could it be that Elon has finally matured to know better when to make bold statements and when to lay low? That could be one possibility, although maturity is not something one gains overnight. Could it be that the Twitter problems he’s in taught him something? After all, if he analyzed everything better instead of jumping head-first into the deal he now wants out of, things would’ve been much better for everybody involved. That could be another possibility.

There could be a third one, too. You know what they say: the more you know, the more you realize that you know nothing. All these advances in the FSD development could’ve taught Tesla’s AI team that cracking FSD is not easy. Although things are improving, a breakthrough will not happen next year. Probably it would not be ready even in the “next 5 to 10 years” when the Optimus Bot should be around to help us pick up the groceries.

If you liked the article, please follow us:  Google News icon Google News Youtube Instagram X (Twitter)
About the author: Cristian Agatie
Cristian Agatie profile photo

After his childhood dream of becoming a "tractor operator" didn't pan out, Cristian turned to journalism, first in print and later moving to online media. His top interests are electric vehicles and new energy solutions.
Full profile

 

Would you like AUTOEVOLUTION to send you notifications?

You will only receive our top stories