Tesla insiders claim CEO Elon Musk had his Full Self-Driving optimized routes that he himself takes as well as routes taken by Tesla FSD content creators, which would explain the discrepancies in the efficacity of the system.
Elon Musk has been praising Tesla’s self-driving effort, and especially its Full Self-Driving (FSD) Beta (now “supervised) system, for years. He is often quoted encouraging people to try it or watch videos of it online to witness how incredible it is.
As someone with FSD Beta form than 2 years, I have often pushed back against this narrative because I’ve found some significant discrepancies between videos from Tesla influencers, mainly in California, Musk’s claimed performance of FSD Beta, and what and my friends have been experiencing in our Tesla vehicles, especially in Quebec and the Northeast.
My suspicion was that Tesla was mostly training its self-driving neural nets on west coast roads, which would explain why it would work marginally better there, but now a new report claims that it goes way deeper than that.
Business Insider released today a new report based on conversations with current and former Tesla workers who claim Tesla has been optimizing its self-driving neural nets specifically for routes taken by CEO Elon Musk and Tesla influencers producing FSD content.
BI spoke with over a dozen current and former Tesla employees, all but one who spoke on condition of anonymity, who said images and video clips from Musk’s Teslas received meticulous scrutiny, while data from high-profile drivers like YouTubers received “VIP” treatment in identifying and addressing issues with the Full Self-Driving software. The result is that Tesla’s Autopilot and FSD software may better navigate routes taken by Musk and other high-profile drivers, making their rides smoother and more straightforward.
The report claims that data annotators, people who review clips from Tesla vehicles to help train its self-driving neural nets, were given the task to focus on two categories: “Tesla CEO Elon Musk and a select set of “VIP” drivers.”
Several annotators who talked to BI confirmed that they were tasked to specifically work on route around Tesla, SpaceX, Twitter locations as well as a mansion previously owned by Musk.
The Tesla employees claimed that this would undoubtedly result in Musk having a better experience with Tesla FSD than the average Tesla driver.
A former Tesla employee said:
“It seemed like we were purposely making his car better to make Autopilot look different than it was. It felt dishonest.”
The report also claims that Tesla focused on routes taken by Tesla FSD influencers. We were already aware that Tesla sends internal test vehicles on routes taken by popular content creators who post FSD Beta videos.
But now the report claims that those were also given priority by data annotators:
These videos do not go unnoticed by Tesla staff. In fact, the company created a system to prioritize data from drivers most likely to share their experience online, three current and former workers with direct knowledge of the issue told BI. These drivers are internally referred to as “VIP” users and their data is at times put in VIP queues, according to the workers.
Some employees claimed that this is due to those content creators pushing the system to theirs limits and therefore, it makes sense to train its neural nets to handle those situations, but it would also results in them having a better experience than the average Tesla FSD Beta driver.
Electrek’s Take
This makes sense. I understand that there are valid arguments for it. Elon is the CEO, but he is also a tester who gives feedback. It would make sense to focus on his feedback. The same goes for the influencers.
However, it also undoubtedly makes Elon’s experience with FSD and the influencer videos not representative of the average customer experience with FSD, which is what I’ve been saying for years.
Where this becomes really problematic is that you can find dozens of comments from Elon telling people to go watch these videos and using that to sell FSD. There’s an argument for misleading promotion here.
What do you think? Let us know in the comment section below.