Elon stated (on the record) that before end of this year, FSD will be "much safer than a human driving" and that he "would be shocked" if FSD does not reach Level 4 autonomous driving is not reached by Dec 31 2022. What say you?
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Post Elon Update Poll on FSD
- Thread starter 2101Guy
- Start date
Mrcarcrazy
In need of a shrinking gun to zap a plaid with.
I’m cautiously optimistic. I figure why not be? I think either outcome is plausible. So I’ll be Optimistic.
diplomat33
Average guy who loves autonomous vehicles
L4 implies a limited ODD and the manufacturer can make the ODD whatever they want and still be L4. So depending on the ODD, sure, Tesla could achieve L4 this year. For example, they could pick a 10 sq mi geofenced area in LA where FSD is able to drive with no human intervention and that would technically be L4. Or Tesla could try FSD that works basically everywhere, except maybe certain intersections that are too tricky. Or Tesla could try do FSD everywhere but limited to good weather. There are lots of ways to do the ODD for L4. It really depends on the ODD. But Elon has really talked about FSD being safer than humans in general. He has not talked about any ODD limits. So I think Elon wants to skip L4 and try for L5. But I think it is likely that Tesla will release FSD later this year that works everywhere but still requires active driver supervision. So it will not be L4 or L5.
Safer than humans I can believe but Level 4 means Tesla will allow no human supervision in some situations. This means that they will be responsible for accidents. I don't see any benefit for them taking on this risk. So no. No level 4 in the near future. They will still require the driver to be responsible for accidents
RTPEV
Active Member
Sorry, I am definitely not seeing the level of capability or progress in FSD Beta to believe that we are anywhere close to this being the case. As much as I admire Elon's gumption and drive to get things done, I really don't think he appreciates the barriers to FSD, and he's demonstrated this since 2016.
I'd generally be okay with his optimistic attitude (it doesn't matter too much to me personally whether FSD becomes reality in 2022 or 2028), except for the fact that his optimism in this area has shifted his (and thus Tesla's) focus away from the $25,000 vehicle in favor of just creating a fleet of autonomous vehicles so that instead of people buying $25K vehicles, they will just hail robotaxis and not own a car at all. I am extremely disappointed at this apparent shift in strategy, not because I don't think that's the eventual point the auto industry will reach, but rather because I believe that point is years (1-2 decades?) out and that in the meantime others are going to swoop in and capture the sub-$30K vehicle market.
I'd generally be okay with his optimistic attitude (it doesn't matter too much to me personally whether FSD becomes reality in 2022 or 2028), except for the fact that his optimism in this area has shifted his (and thus Tesla's) focus away from the $25,000 vehicle in favor of just creating a fleet of autonomous vehicles so that instead of people buying $25K vehicles, they will just hail robotaxis and not own a car at all. I am extremely disappointed at this apparent shift in strategy, not because I don't think that's the eventual point the auto industry will reach, but rather because I believe that point is years (1-2 decades?) out and that in the meantime others are going to swoop in and capture the sub-$30K vehicle market.
The red part: I can SO see Elon pulling that move in Q4 of this year when pressed on it. "I never said L4 anywhere and everywhere on the globe. I merely referenced the capability being achievable at scale in an exponential manner within a specified radius. And doing so would dramatically support the push toward robotaxi service, a paradigm which would exponentially raise the value of a tesla by margins previously unheard of".
Or some other nonsensical babble.
Elon’s whole view around FSD does seem
kinda strange nowadays. 100% of the point here is to increase vehicle utilization and have Teslas out operating as Robotaxis rather than sitting in a parking lot. Somehow this will be super cost effective for users, less than the cost of a subsidized bus fee, while also making sense economically for the owner of the vehicle. I don’t quite know how that will work out or how it meshes with the idea that FSD will become far more expensive over time.
I think Elon does believe what he‘s saying but that what he’s saying isn’t aligned with peoples
expectations. Elon thinks this year they “will achieve Full Self-driving” to a level safer than a human driver, which he says is a very low bar and that getting to the point of being 1000-10,000% better than humans will be the real challenge.
I can see than producing some statistic like the Autopilot safety stat by the end of the year. But all this said, it won’t be Level 4/5 and it won’t be Robotaxis.
It was also a bit disheartening to hear that Dojo really is just a tool for cost effectiveness and won’t bring a new wave of exponential improvements as I think many believed
kinda strange nowadays. 100% of the point here is to increase vehicle utilization and have Teslas out operating as Robotaxis rather than sitting in a parking lot. Somehow this will be super cost effective for users, less than the cost of a subsidized bus fee, while also making sense economically for the owner of the vehicle. I don’t quite know how that will work out or how it meshes with the idea that FSD will become far more expensive over time.
I think Elon does believe what he‘s saying but that what he’s saying isn’t aligned with peoples
expectations. Elon thinks this year they “will achieve Full Self-driving” to a level safer than a human driver, which he says is a very low bar and that getting to the point of being 1000-10,000% better than humans will be the real challenge.
I can see than producing some statistic like the Autopilot safety stat by the end of the year. But all this said, it won’t be Level 4/5 and it won’t be Robotaxis.
It was also a bit disheartening to hear that Dojo really is just a tool for cost effectiveness and won’t bring a new wave of exponential improvements as I think many believed
I'm pretty sure this is the 3rd year i've heard him say this. I believe he's losing touch with the realities of real world human driving. LIstening to the earnings call, i heard a company and a CEO obsessed with autonomy and robotaxis. Whenever i hear any of them talk, it tells me that their end goal is to stop all manual driving. I wonder why exactly this would be. If the answer is for the safety of humanity, i don't buy it.
When a global figure such as Elon, repeats something over and over for years, it starts to imitate some great truth or fact. It started before him, but this great truth is that autonomy is the future. No one in the future will drive themselves, except a few enthusiasts, like those that own and ride horses. I call bullshit on this.
Thankfully, i don't believe FSD will be much safer than humans by the end of the year. If that ever does happen, we are in danger of losing another liberty. Not that it will be taken away explicitly, but Tesla insurance or something like it will be required. We will certainly be free to take manual control of the car, but our insurance bill will reflect the number of times we do that (and also how we drive on manual override). We will be become prisoners of autonomy by default, unless we are wealthy enough not to care about our insurance bill.
I've been running FSD since 10.5. While 10.9 is a vast improvement over 10.5 it still seems a stretch to get to full autonomy to the point where you have enough confidence in the car that you can do other things in the car while the car is driving. I can however see FSD improving to a point where 90% of the drive the driver can be inattentive to the actual physical exertion of driving while still paying full attention to the road (Like how AP is on highways right now).
Many seems to think that FSD value is all or nothing, that it is only of value if L4 FSD is fully achieve. Personally, I think there is still a huge value to FSD when FSD can take over majority of the task of driving and leaving the driver to be "in command" of the vehicle like the captain of a ship. On my daily commute using FSD, we are basically already there now with 10.9. The lion share of my disengagement these days are related to stupid lane choices, and to tell the car it's safe to make the turn/intersection sooner than it can make that determination. These aren't safety type disengagement but "don't be a jerk on the road" type of disengagement. I can see safety type disengagement going to ~0 by the end of the year while the "don't be a jerk" type of disengagement still taking many more years to resolve. Both needs to be resolved before full L4 FSD is achieved in my book, which is why I voted negative in the poll. But by the end of the year I do think Tesla FSD will be at a point where there won't be safety type of disengagement and will become a legit useful driving aid.
Many seems to think that FSD value is all or nothing, that it is only of value if L4 FSD is fully achieve. Personally, I think there is still a huge value to FSD when FSD can take over majority of the task of driving and leaving the driver to be "in command" of the vehicle like the captain of a ship. On my daily commute using FSD, we are basically already there now with 10.9. The lion share of my disengagement these days are related to stupid lane choices, and to tell the car it's safe to make the turn/intersection sooner than it can make that determination. These aren't safety type disengagement but "don't be a jerk on the road" type of disengagement. I can see safety type disengagement going to ~0 by the end of the year while the "don't be a jerk" type of disengagement still taking many more years to resolve. Both needs to be resolved before full L4 FSD is achieved in my book, which is why I voted negative in the poll. But by the end of the year I do think Tesla FSD will be at a point where there won't be safety type of disengagement and will become a legit useful driving aid.
K1Woods
Member
I think you bring up a good point about insurance but in my case a 30% savings on my Model Y insurance is less than $300 so that isn't enough at least for me to worry about. It's also interesting that teenagers today don't nearly have the same passion for driving as the baby boomer generation did so many people won't share you concern about not driving. I for one wonder if my young grandchildren (ages 2-8) will drive much anyway especially with both families also owning Teslas and the grandkids have seen my FSD in action. (simple drives only)
Teens Don't Have Interest in Driving Anymore and That's OK
URBAN LEGEND
Doing it well
But if a FSD system is better than the average driver, that insurance is going to cost way less than it does today.
As individuals, we think we are good drivers. But as a society, the numbers point to a different picture. If I imagine an FSD system that is better than society's average, say 10x better, I can totally see a world where driving is mostly autonomous, and people won't perceive it as a loss of liberty. It will be seen as a convenience and an improvement to society. I don't consider it a loss of liberty of having to manually hand crank a pump to get water from my well.
Now, as for Tesla's FSD system.... I think even if at the end of 2022 FSD has matched the safety of human drivers, or is even 2x better, public perception will not be swayed. We are just not very forgiving of machine-type mistakes. I think it has to be better than 10x the societal average, and even then, it will take some convincing that the system is actually >10x better, since the stats can always be distorted to push various narratives, good or bad. Kinda like current AP.
RTPEV
Active Member
My primary complaint with FSD Beta (and the reason I consider it 100% useless to me at this point) is how it handles uncontrolled (and by that I mean things like 4-way stops, T-intersections, and roundabouts) intersections. I am wondering what your experience is, and if it's just me and any unique characteristics about the places where I drive.
My experience is that when approaching such an intersection the car will stop (good), but then creep every so slowly up to the edge of and well into the intersection, going at 2-3 mph the whole time. This is not only annoying and confusing to other drivers at the intersection, but I believe dangerous as the expectation is that once the car enters the intersection, the expectation is that it will accelerate out of it, not continue to creep through it. Since I have to go through several of these types of intersections each day, I basically find FSD beta useless, and I haven't seen any progress (and maybe even a little regression) since I first got FSD beta in late October (granted, there haven't been that many updates since then). And I'd call this relatively low hanging fruit.
I do continue to engage when there are no other cars around just so I can monitor progress, but I haven't seen any.
I do agree with you that a L4 FSD could be valuable. For example FSD doesn't handle my neighborhood well at all: no center lane markings means the car basically centers itself in the middle of the road (even though it's as wide as a two lane road and the car should realize it should stay to the right), and this becomes a huge problem when it has to turn left onto my street--it cuts the corner severely, hugging the left side of the street which could be very dangerous. But if that's the only place it didn't work, then fine...I'll just take over as I approach my neighborhood and enjoy FSD the rest of the time. But as in the example above, there are so many places where it just fails to operate correctly that it's really not worth my time as I'm constantly engaging and disengaging it anyway. And yet I see comments from YouTubers and Elon himself basically praising current revisions of FSD Beta as being mind blowing an no disengagements whatsoever, and it leaves me to wonder what kinds of roads is this being tested on? Are the only intersections being encountered controlled by stoplights?
When did he say level 4? They aren’t even working on level 4…
RTPEV
Active Member
I think the point that @hatch is making (and it's an excellent one) is that for those who do not want autonomous driving, or cannot (because their typical trips may not be conducive to autonomous driving, or they live in a very rural area with insufficient density of robotaxis), their insurance rates (because they are not using autonomous driving) may increase significantly. The way I see it working out is that insurance rates will not actually go down (or at least not much), but as the shift to autonomous driving happens, premiums for those who do not take advantage of autonomous driving will skyrocket.
It's impossible to know what will happen of course, but I see this as at least a possibility.
Wooloomooloo
Member
Should I dig up my posts from 2018 and 2019 (heavily down-voted I might add) where I said FSD Level 3 would not arrive in those years, let alone level 4 many people were arguing for! 
In all seriousness, FSD will continue to improve incrementally for the next decade or so. Level 4 is quite a way off and likely not possible with the current cameras and compute available in today's Model 3's and Y's. You'll probably find most people who have FSD (I do) will vote no, whereas most people who don't even have a Tesla, just a poster of Elon on their bedroom wall, will vote yes.
**edit - someone already pointed out that Musk hasn't even mentioned Level 4, and likely for very good reason.
In all seriousness, FSD will continue to improve incrementally for the next decade or so. Level 4 is quite a way off and likely not possible with the current cameras and compute available in today's Model 3's and Y's. You'll probably find most people who have FSD (I do) will vote no, whereas most people who don't even have a Tesla, just a poster of Elon on their bedroom wall, will vote yes.
**edit - someone already pointed out that Musk hasn't even mentioned Level 4, and likely for very good reason.
Similar threads
