Fully agree and that’s been my assessment using 10.5 so far. Perfect no but pretty darn close. Much closer then anyone just now getting into this technology. As for the beta call out, fully agree. They will keep that definition for a Long time while many are trying to fault the system. A safety net of sorts.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
GM’s new Ultra Cruise: Hands-free driving on all paved roads in US/Canada
- Thread starter stcptl
- Start date
EVNow
Well-Known Member
Yes - thats why I qualify FSD Beta as "early access". When I say general availability - I don't mean in traditional sense, but like how NOA is available to everyone who pays.
My crystal ball, which is notoriously murky, says...
Elon is always a few moves ahead of everyone else. He has orchestrated this plan of releasing Beta FSD on a wider basis because Tesla is rapidly approaching the base of the exponential improvement curve. Dojo (and the other tools needed) to rapidly improve the capabilities of FSD are darn near ready to go.
So now that the testing group has grown to its current level, Tesla has accomplished a couple of things: 1. It's now well known how good/awful FSD Beta is in its current state. 2. They have a fleet setup to "feed me, Seymore" when Dojo is ready to start crunching data. 3. It gives them time to gather, label, and QC a huge amount of data to get Dojo off to a flying start.
We're going to see another few months of underwhelming releases. But then, we'll start to see very noticeable improvements with every bi-weekly release as Dojo starts throwing its muscle at the problem. We'll also see a lot of hard coded things being switched over to decision based.
Assuming that Dojo comes online in the spring of '22, by this time next year the FSD Beta we'll be testing won't even seem like the same software we're currently running; we'll be well along the exponential improvement curve. The rate of improvement will even be good enough for Gordon Johnson to buy Tesla stock... OK, maybe not to that extreme.
We will be looking back at October of '21 saying, "well played, Elon.. well played."
At this point, my crystal ball becomes too murky to read.
Prior to watching the AI Day presentation, my crystal ball said "we're at least five years out." It has since changed its tune to L3 by the end of next year.
Regardless of whether I'm right or wrong, it's going to be one hell of a fun ride to be on.
And if you still haven't watched the AI Day presentation, GO WATCH IT!!! Yeah, it's a bit over three hours long. Eat that elephant in small bites if you have to, but eat that elephant! So much of the stuff the nay sayers are saying on this board is covered in depth in the AI Day presentation. And it's presented in a clear cut, "here's your answer" format; no hand wavy crap.
Elon is always a few moves ahead of everyone else. He has orchestrated this plan of releasing Beta FSD on a wider basis because Tesla is rapidly approaching the base of the exponential improvement curve. Dojo (and the other tools needed) to rapidly improve the capabilities of FSD are darn near ready to go.
So now that the testing group has grown to its current level, Tesla has accomplished a couple of things: 1. It's now well known how good/awful FSD Beta is in its current state. 2. They have a fleet setup to "feed me, Seymore" when Dojo is ready to start crunching data. 3. It gives them time to gather, label, and QC a huge amount of data to get Dojo off to a flying start.
We're going to see another few months of underwhelming releases. But then, we'll start to see very noticeable improvements with every bi-weekly release as Dojo starts throwing its muscle at the problem. We'll also see a lot of hard coded things being switched over to decision based.
Assuming that Dojo comes online in the spring of '22, by this time next year the FSD Beta we'll be testing won't even seem like the same software we're currently running; we'll be well along the exponential improvement curve. The rate of improvement will even be good enough for Gordon Johnson to buy Tesla stock... OK, maybe not to that extreme.
We will be looking back at October of '21 saying, "well played, Elon.. well played."
At this point, my crystal ball becomes too murky to read.
Prior to watching the AI Day presentation, my crystal ball said "we're at least five years out." It has since changed its tune to L3 by the end of next year.
Regardless of whether I'm right or wrong, it's going to be one hell of a fun ride to be on.
And if you still haven't watched the AI Day presentation, GO WATCH IT!!! Yeah, it's a bit over three hours long. Eat that elephant in small bites if you have to, but eat that elephant! So much of the stuff the nay sayers are saying on this board is covered in depth in the AI Day presentation. And it's presented in a clear cut, "here's your answer" format; no hand wavy crap.
EVNow
Well-Known Member
LOL. The reason Gordon will never have a Buy on TSLA is that the stock will always be ahead of him.Assuming that Dojo comes online in the spring of '22, by this time next year the FSD Beta we'll be testing won't even seem like the same software we're currently running; we'll be well along the exponential improvement curve. The rate of improvement will even be good enough for Gordon Johnson to buy Tesla stock... OK, maybe not to that extreme.
BTW, when they start using DoJo, we'll be back to square one with all the regressions like in VO vs radar
As I said, an optimistic estimate is 2 years for GA.
Clear and concise... well done.
On a certain level, I even agree with some of the points you've made.
What I do kinda take issue with is number 5. I just don't think that, as lay people, we have a good enough understanding of the abilities and limitations of the various sensor options to really be able to form a valid opinion. After all, Elon said that SpaceX does use lidar to dock Dragon with the ISS; it's very good at what it can do, but it does have its limitations... just as all sensors do.
At one point, one of the guys at Tesla was asked why they were removing radar. After explaining it on a technical level, he also explained it for dummies like me; "using radar for this problem is like trying to see with your ears."
But yes, no matter what sensor(s) ends up being the Holy Grail for FSD, some way of keeping it/them clear, clean, and unobstructed truly seems like a no-brainer.
The reason you think Elon was the only one overhyping autonomy is because of skewed media coverage. If you dig, you will see that the entities you purport to trust more, like Waymo and plenty others, also underestimated the time required to achieve autonomy.
Why we still don’t have self-driving cars on the roads in 2021
The promise of self-driving cars hasn’t materialized, and we’re no closer to them becoming a reality.
theconversation.com
Ford CEO says the company 'overestimated' self-driving cars
Ford CEO Jim Hackett scaled back hopes about the company's plans for self-driving cars this week, admitting that the first vehicles will have limits. "We overestimated the arrival of autonomous vehicles," said Hackett, who once headed the company's autonomous vehicle division, at a Detroit...
www.engadget.com
For years, automakers wildly overpromised on self-driving cars and electric vehicles—what now?
Tech execs and auto company leaders are becoming increasingly pessimistic about the future of AVs—and they are far less sanguine about electric vehicles (EVs) as well—because the raft of rosy projections in recent years have not even remotely come true.
Also, what you might not realize is that Waymo's strategy for going with multiple high-cost sensors means very large initial capital expenses, and it's unclear if demand for their services can cover those expenses. So interestingly, they may be closer to autonomy, but their approach may not be economically feasible.
Subjectively, I like Tesla's pure vision, heavy AI methodology because they are pushing AI tech forward, which will have way more applications than just AVs.
Haha that’s so accurate. Radar is like seeing with your ears! To that I would have responded, but sometimes when you can’t see something approaching, perhaps your view is momentarily blocked, you CAN hear it approaching, and you know to run away. Or hit the brakes
The reason you think Elon was the only one overhyping autonomy is because of skewed media coverage. If you dig, you will see that the entities you purport to trust more, like Waymo and plenty others, also underestimated the time required to achieve autonomy.
Why we still don’t have self-driving cars on the roads in 2021
The promise of self-driving cars hasn’t materialized, and we’re no closer to them becoming a reality.
Ford CEO says the company 'overestimated' self-driving cars
Ford CEO Jim Hackett scaled back hopes about the company's plans for self-driving cars this week, admitting that the first vehicles will have limits. "We overestimated the arrival of autonomous vehicles," said Hackett, who once headed the company's autonomous vehicle division, at a Detroit...
For years, automakers wildly overpromised on self-driving cars and electric vehicles—what now?
Tech execs and auto company leaders are becoming increasingly pessimistic about the future of AVs—and they are far less sanguine about electric vehicles (EVs) as well—because the raft of rosy projections in recent years have not even remotely come true.www.fastcompany.com
Also, what you might not realize is that Waymo's strategy for going with multiple high-cost sensors means very large initial capital expenses, and it's unclear if demand for their services can cover those expenses. So interestingly, they may be closer to autonomy, but their approach may not be economically feasible.
Subjectively, I like Tesla's pure vision, heavy AI methodology because they are pushing AI tech forward, which will have way more applications than just AVs.
People made inaccurate predictions. No one was foolish enough to sell it in advance by the millions.
Cost must have come down enough that radar is ubiquitous and LiDAR is starting to appear on production vehicles.
IMHO the correct way to look at all this is to ignore every single timeline prediction from all companies. This is all trail-blazing technology, regardless of the arguments over radar vs lidar vs cameras, or L2 vs L5 or L-anything. No-one has a CLUE how long it will take, or which technology stack or approach will yield the best results (or any at all) long-term.Funny how other companies can underestimate the L5 timeline and not get called to task, but when Elon does it...
I typed out a rather long rant here, but decided to delete it all. No reason to add additional stirring to this self-stirring pot!
But watching (and taking part) in it happening is damn exciting as far as I am concerned.
I'm in that camp for sure. Its frustrating but equally interesting seeing what improves and what doesn't.
If anything I'm more convinced that Tesla will make it work, but I'm also equally convinced it won't be anytime in the next two years (maybe three)
Its so difficult to predict, with releases happening so rapidly most improvements are bound to be incremental and almost unnoticed. It will be fascinating to compare 10.5 against whatever is released one year from now.
I've seen the AP that was delivered when I got my car be transformed from the timid and slightly scary AP to the confident and comforting AP that was my car before I got FSD beta.
Sadly now I'm back at the timid and slightly scary stage again
But hey, in two/three years maybe it really will be confident and comforting again?
The only reason I don't trust GM is that I just don't trust GM with anything. They will find a way to make it useless, they always do.
EVNow
Well-Known Member
FTFY
The truth is, complicated, though.
All others had deep pockets (or deep pockets backing them). Tesla was the only one trying to do two "impossible" things at the same time - EVs and AVs.
BTW, remember all those "not foolish" people thought you can't make a profitable EV, let alone make them best sellers. They thought it is better to hype Hydrogen - I bet you believed they will have a lot of hydrogen cars running around everywhere by now.
Tesla sold AP because they firmly believe it improves safety. Musk has said many times that it would be morally irresponsible not to make it available to as many people as possible. Now, whether the system (current production AP/FSD) is actually safer is the subject of debate. But I'd hardly call Tesla's motivations foolish. The intent is to constantly improve the safety of these systems so that it becomes indisputable that it is safer.
Blue Cruise and Supercruise are in the hands of real people as well. How is Tesla alone? Munro tested Blue Cruise and it couldn't handle a gentle turn on a highway.
BlueCruise should’ve worked from the start but highway is easily fixable OTA, but FSD is not easy. Musk has been selling not just AP but also FSD and L5 for years. How many of those people have already sold their cars? That is quite possibly illegal.
No one has been as careless (public safety & AV public image, possible illegality) and wrong with FSD as Musk has. There are NHTSA investigations going on regarding this.
Oh come on! Virtually every car maker ever has fought tooth and nail to keep major defects from being publicly revealed for decades, sometimes with disastrous consequences for the public. How can you possibly say Tesla are the worst? Based on what? That you dont like Elon?
EVNow
Well-Known Member
You got to be kidding me. Legacy automakers have bought over NHTSA and for years hidden defects that killed a lot of people just to save the companies a few cents per car. You are fast losing any credibility here with such obviously incorrect statements.
ps : I see you are less than 2 months on this forum and you are in Toronto. Are you long on GM or employed by a legacy automaker ?
Similar threads
- Article
