Confronting MeetKevin on FSD v12
ฝัง
- เผยแพร่เมื่อ 4 เม.ย. 2024
- MeetKevin uploaded a video today, and one of the comments he made was to do with his lackluster experience with FSD v12... Suggesting that v12 may even be a temporary step back from v11.
Well. I couldn't disagree more. Here I explain my reasoning. - วิทยาศาสตร์และเทคโนโลยี
why do people care what Kevin has to say? he's a goof, not a Tesla, AI or FSD expert.
Isn't that his profession, influencer? Lot's of ppls money are long/short for the stock.
I think what is more important to look at is the state of v12 today, but the rate of improvement it has shown since its release.
and the crazy thing is FSD v12 is just the "will smith eating spaghetti" equivalent of self driving AI. This thing has been training for like a year, and only now is it being widely pushed out to the masses. From the data and training it will be collecting in the coming 1-2 years, it is gonna go from where it already is now, to making a leap like will smith spaghetti to OpenAI Sora.
My understanding is that Meet Kevin sold his TSLA stock to tax loss harvest and is planning to buy it back when the period is up.
Risky move.
excellent argument, well presented and with respect. well done!
FSD V12.3 is without a doubt a huge improvement. Kevin is in the extreme minority of those of us with FSD on this one.
I am 100% on you with that. MeetKevin is a smart guy with a big flaw of overestimating his capabilities and make tons of assumptions which he then demonstrates as facts. Probably because of being a real estate agent, he tries to sell his thoughts even if he lacks the know how....
Kevin IS Awesome! Even if he doesn't like his Tesla. I am a Tesla Guy. The reviewer and Kevin are Fantastic speakers! Thank you Both !!!
Kevin clearly doesn’t drive enough to see the MASSIVE difference between v11 and 12. The way v11 still acts funny on the hwy and can’t make intuitive decisions, and how shakey and scared it was is palpable. V12 is an extreme jump forward. It has solved literally 19/20 bugs v11 has absolutely disappeared. This is demonstrably the correct path.
1B miles FSD driven until march2024
When 6B are reached it will be perfect !
Subscribed!
i love kevin but i think he's wrong about v12... however with that said, he has a model X, from what i have seen so far it seems like model S and X v12 is less consistent than V12 is on 3 and Y for some reason, maybe because there are way more data on 3/Y than S/X since the fleet size is larger
v12 is great on my 3 though, great video! i think V12 has some navigational issues and it still struggles on parking lots but it is definitely a quantum leap from v11
I agree with you 100% version 12 is leaps and bounds ahead “paradigm shift“. I have been driving my model three 2018 enhanced auto pilot for over 80,000 miles adding years of data and bug report and then my 2023 model Y enhanced auto pilot now driving in self driving beta For over a year I affirm that version 12 is a paradigm shift. It’s time to get Robo taxi on the road. It’s time to get beta testers to allow their vehicles to drive by themselves. Autonomous beta we should be ready soon is what I say before the end of the year.
great video
Top tip... "K" likely won't improve your numbers. 😉
Keven !!!
meetKevin is now ShitKevin
12.3.3 feels better in my wifes M3 with HW3 than it does in my MYP with H4. Its of course subjective.
Meat Kevin...
12 vs 11...... Night va Day
It's still not a ready product. It had disengagements with other youtubers.
Depends on what you mean by "product." Certainly it's not robotaxi level yet. But it is now good enough that many would find it worthy of some price to handle most of their driving.
@@DeusExTesla You can use your phone on mercedes one. I know it's worse but you can use it legally, but you have to have hands on wheel on fsd.
I don't have my hands on the wheel 99% of the time with FSD, it's really quite reliable.
hi is Moran and I don't know why everyone is lessening to his BS
Wth is kevin talking about?
Yeah, no one gives a fig what Kevin thinks or doesn't think or whatever
FSD v12 is FSD AI v1.
Give it 10 more versions and then compare FSD v11 to FSD AI v11.
You're using anecdotal evidence to try and disprove anecdotal evidence.
friendly reminder: Kevin hyped the stock of Matterport for a good while. the stock lost 80 Percent of its value over the last five years, because they basically sell a joke of a product. He is a joke and deceptive. Dont trust him.
Lets be honest.... v12 is shockingly fantastic, except for when it's NOT.... and it's NOT very good at lane merges. Tesla should have waited for wide release of v12 until this most basic function was solved. .4 is not any better than .1, so we are back to the same old bullshit.
I stopped watching kevin months ago. He is like a used car salesman.
meet kevin is and has always been a grifter. He's betting on Tesla going down so he can buy and then he'll start preaching again. Just watch.
100%
FSD relies on clearly-painted demarcations on the roads. I'd bet people living in cities with better-maintained, better-painted roads have better experiences with FSD, than people who live in cities which skimp on road maintenance. Many American cities are run by notoriously neglectful governments.
I'm not a fan of Kevin but you are actually wrong here. v12 is very human like in its smoothness and I get that seems like a huge shift but it makes many of the old types of mistakes and some new ones. clipping curbs, missing lanes, confusing red lights, crossing red lights, driving into booms, missing speed limits etc and that 12.3.3. It might be genuinely better at road works but overall for taxi use, level 4 autonomy, it's not a decisive difference. Kevin might be a little callous but he's not substantially wrong here.
Have you used FSD 11 and 12 substantially? From my experience so far, I would feel dishonest to suggest it's not anything but a massive and transformational leap. It's better in every conceivable way. Makes mistakes, yes, but nowhere near as much.
@@DeusExTesla no I've only watched a lot of test videos of both. have you had it head straight for a curb? not just graze. Certain roads of acute angles will also miss. Everybody is saying it drives like a person which makes sense because it is mimicking people but no one is saying autonomy is imminent. Even Mars catalog.
@@DeusExTesla - I have used both substantially. At least 25 hours on v11 and around 6 on v12.3.3 now. There are some regressions. I've had it turn right on red when a sign clearly said no. I've had it shit lanes multiple times and end up stopping hovering over the white line. The biggest issue for me though is at 1 way stops pulling out into fast moving cross traffic. The car is just flat out indecisive and while it "may" not actually wreck it scares the shit out of me when it clearly has room to make it, but then pauses and waits until to act like it will go again. Go or don't go. Decide then execute. I hate the 'thinking' and watching the wheel move and the car go/stop...no go/stop...
But of course you are right about other things. I took it into downtown Boston today and was pretty well blow away 90% of the time. Very hard to drive in Boston, but the other 10% it was not so great.
@@DanFrederiksen I did indeed just have it almost drive into a curb today. That was the first serious/scary intervention I've had with it in over 2 weeks of using it. It was a strange angle, as you said, so it was a pretty rare scenario.
I still think we could be only a year away from virtually robotaxi level driving. It seems to me, they just need to add more data for more edge cases, and they should be able to access that data pretty easily.
We will see.
@@johnfurr6060 Yeah, agreed. Some of those cross-traffic turns can be too hesitant and it ends up sitting in the middle of the street. I feel confident they can improve those though, based on the other improvements i've seen from v12.
FSD V11 sucked at self driving
FSD V12 sucks less
That's all you need to know.
Exactly - it’s still not perfect or ready for “robo taxi” launch - Kevin even mentions this several times
@@Eddy-ov4tx FSD is doomed to failure - he probably didn't get into that.
It is based on neural networks, which is probably the worst way to write a self-driving app.
And FSD is unable to do its primary task - drive folks around.
NO state will certify it as self-driving since the ability of a police officer to test it
was never even considered.
FSD's primary goal is to become your Chauffeur, which means it must pass the Chauffeur's driver's license test in the various states. That means a police officer sits in the car and tells FSD where to drive, when to stop, and when to park. FSD is deaf; it can't hear.
Elon hasn't spent 5 seconds on how FSD is to be certified in the USA and other countries.
I could go on and on, but FSD is doomed to failure, and to those of us who paid $10,000 to get it, we will never get what we paid for - a Chauffeur...
Driving is a privilege granted by the state - and the state has been left out of FSD from day one.
California will be the first to demand FSD be geofenced and not active on California roads - just you watch.