⭐️ FSD and robotaxi are coming.

In this second quarter’s Tesla earnings call, Robotaxi Unveil has been postponed to Oct. 10. There are various reactions, but here’s what I think.

• The public response to FSD and Robotaxi is still at the Lidar or Camera level (please…)
• Tesla Elon? Fsd has been doing this year for years, and you’re lying again, and you’re tricking me again?
• There’s a lot of reactions to robo-taxi acting, from “I knew it” to “I can’t do that because of the regulations.”

However, the driving data of real users who use FSD every day in the United States are as follows.
(See attached image)

Total mileage: 4866 miles
FSD mileage: 4740 miles
FSD driving ratio during overall driving: 97.4%

Many of the remaining 2.6% of the driving are last mile and early parking lights.

FSD in a faraway country is the only one in the U.S., right? In the meantime, there was an accident that happened in the early days of Autopilot in the past, and there was a debate many years ago about whether the term FSD was right, while you were still stopped
FSD is doing a big jump like AI LLM model is updated every time one version is updated.

When you talk about regulatory hurdles,

Tesla’s regulators are already discussing deregulation and licensing with those responsible for the policies of U.S. presidential candidates, regulators in Europe and China, and real powers.

• Many people will know that the recent rapid development of AI is so distracting.
• But there is a lot of question these days, “What do you want to make money from it?”
• I’ve been thinking about one thing for a few years. It’s the most obvious thing to do, and it’s going to be the self-driving, the robotaxi, the humanoid, that’s going to make a lot of money with AI
• In FSD V12.5, the parameters of the E2E AI model are extended by 5x compared to the previous one.
• We’re seeing huge improvements from the early versions, not the debugging versions. Perhaps we’re expecting another free trial version distribution from the debugging versions of Highway OnceTack.

✅ People are mistaken. I think self-driving is a skill, and I think we should develop self-driving completely and safely before releasing it to the world.

✅ However, autonomous driving is the kind of thing that can never be done in the first place. It’s a tedious process of developing that keeps increasing the number of 9s: 90%, 95%, 99%, 99.99%, 99.999%, 99.999%.

Elon Musk and Tesla’s AI team have recently talked about the difficulty of collecting data as FSD has developed and human intervention has been significantly reduced. No, it is difficult to collect data even at Tesla, which actually has the most real-world data? What does this mean? You’ll be wondering.

With the development of FSD, 97 percent of the driving data is literally good, but it’s become meaningless. It’s getting harder and harder to get good edge case data.
(Of course, in reality, 97 percent of the data should be fully learned.)

It’s actually the only company on the planet that collects real-life data using billions of miles of autonomous driving already making this appeal…
I’m better because I’m level 3 without running 1,000 units in a specific area yet?
(If you look at the level 3 support Mercedes that I’ve posted before, all I can do is laugh. https://www.facebook.com/share/p/HPMgyGePB4gFaJQx/ ?mibextid=oFDknk)
I can only think of it as a dog noise.

✅ Then, you may be asking the question, “Who will be responsible for what happens in the process that is not perfect?” I agree with some of the various opinions on that and respect those opinions. However, as I mentioned earlier, there can never be a perfect driving for autonomous driving. The only way is to increase the number of continuous 9s at level 2 despite the controversy. And even though the manufacturer is not perfect when it reaches a certain level, it will be a true level 3 autonomous driving, not the current ridiculous fake level 3 as rain when it reaches a certain level.

It’s a process of constantly pursuing perfection. The connection between 6G and everything? I bet it’s hard even if it takes 100 years.

✅ Currently, domestic and European regulations have manufacturer responsibility hurdles from level 3. Level 2 imposes too many restrictions to impose autonomy. For example, to change lanes, it is essential to verify the operation of the user, and to finish the lane change within a few seconds. So DACS emerged. But even this has many restrictions. (Even that DACS in Korea…) It is asserted that regulations that impose level 3 responsibility and stifle level 2 autonomy are the worst regulations that stand in the way of development.

🔥 Elon spoke on an earnings call. Sell Tesla stock if you don’t believe FSD is going to be done. The rest is just noise. I’ve been believing for years. This is the only company that will do this if it can commercialize self-driving.

⚠모델자주보는오토입니다의Y최근 ️’s Model Y This is a recent review video. It’s a little strange that you just posted an old car, but… What’s more surprising is that you’re surprised by the basic Autopilot performance… You’ve driven a lot of cars in that test section… I’m really curious about your reaction if you do FSD.
I edited and uploaded some of them, but if it’s a problem, I’ll delete them.

🎶 Some people say that there is not much difference between Autopilot and other ADAS technologies. They have never used Autopilot and its self-driving technologies to the limit. As I mentioned earlier (although Autopilot is not an autonomous driving technology), these technologies are 99% or 99.99% when they reach a certain level. You have to try Autopilot to the limit so that you can understand the limitations clearly and use it more safely.