Tesla has uploaded a new video demonstration of the newest Full Self-Driving Beta software recorded on the streets of Austin, Texas.
The five-minute long video posted on Tesla's official X (formerly Twitter) account aims to show how FSD uses "vision neural networks to perceive & understand the world, just like humans do." You may recall that in late August, Tesla CEO Elon Musk did a livestream of a drive on the company's latest FSD software – V12 – in the Bay Area.
While Musk's video also focused on how FSD works as a human brain, using neural nets and "eyes," the footage was rather underwhelming because of poor filming and image quality issues. The FSD system also experienced several hiccups during Musk's livestream drive, which was 45 minutes long.
Tesla's latest video is clearly of higher visual quality, albeit much shorter. While the automaker doesn't mention the exact version of FSD Beta tested in the video, it does say it is "version 11.4.7 or later." This is somewhat strange because it's not clear if Tesla used the latest neural network-based vehicle controls in this demonstration – that capability is said to have debuted on FSD Beta V12.
It's also worth noting that Tesla said it removed the alerts to apply pressure on the steering wheel for this demo, which is something that regular owners aren't able to do. Tesla likely did this to make the video look better, but that's not the way FSD Beta is supposed to be used.
With that out of the way, FSD Beta appears to do a great job of navigating all sorts of scenarios, from dealing with stop signs at intersections in suburban areas, stop lights at larger intersections, and the infamous unprotected left turn.
The highway stuff is something that FSD Beta knows best so there are no surprises there, while the latter part of the demo sees the Model S cruising through downtown traffic before reaching the destination in the suburbs.
Should You Believe Tesla This Time?
This is all good, but we must remember that Tesla staged its 2016 self-driving demo video, according to a July 2022 court testimony from Ashok Elluswamy, director of Tesla's Autopilot Software.
Elluswamy said the system did not have the capabilities shown in the video at the time, such as stopping at a red light and accelerating at a green light. He also noted that Tesla used 3D mapping on a predetermined route from a house in Menlo Park, California, to Tesla's then-headquarters in Palo Alto, adding that drivers intervened to take control in test runs.
Despite that, the video (see it below) was promoted by Elon Musk on Twitter in 2016 as evidence "Tesla drives itself."
Why would Tesla go to such great lengths to convince people its FSD tech can do things that it actually cannot? It has a lot to do with the fact Elon Musk is tying the future of Tesla's market valuation to FSD.
Last year, he said solving self-driving is the difference between Tesla being worth a lot or nothing. "The overwhelming focus is on solving full self-driving. That's essential. It's really the difference between Tesla being worth a lot of money or worth basically zero," Musk said.
Then in January 2023, he remarked during an earnings call that only some "smart retail investors" see the real value of Tesla's FSD.
“Something that I think some of these smart retail investors understand but I think a lot of others maybe don't – is that every time we sell a car, it has the ability, just from uploading software, to have full self-driving enabled, and full self-driving is obviously getting better very rapidly," Musk said.
The real question is when will Tesla vehicles be capable of actual "full self-driving," whatever Musk understands by that. According to comments he made in April, Tesla may achieve "full autonomy" this year. The only problem with that is he also said self-driving technology would be ready and widely available by the end of 2022.