We'll know Tesla is serious about robotaxis when it starts hiring remote operators.
See full article...
See full article...
On self-driving, Waymo is playing chess while Tesla plays checkers
- Thread starter JournalBot
- Start date
Agreed, it shouldn't be a Tesla-bashing just because Elon is at the helm and he's On the Other Team, but their technology is just incredibly immature and, like many things at the company, is a strong bet on future, wildly-improved technology on which no one has any vague idea of timeline.
Humans can operate without road signs because we have an insane amount of training data and far more reliable intelligence than the state of the art and far better visual perception; there's no reason to believe that AI will get to that level of competency in the next decade even (other than the amount of cash being dumped into it). But unless both that and insane, eyeball-matching cheap camera hardware come to market in the next year, Tesla is years behind. Much more likely is that they are over-promising and will under-deliver, just as they have for the past several years on the topic. They're also just definitionally years behind, since they don't have any driverless cars on the road and regulators want to see a ton of data over time before widening permits. Actually full self-driving is much harder to roll out, logistically, than what they've been doing so far.
These autonomous drivers need to not just match human levels of reliability and safety, but significantly surpass it in order to appease the public and regulators. If Tesla continues down this path, the most likely outcome I see is either getting themselves or the whole industry blocked with legal impediments for years.
Last edited:
Upvote
22
(25
/
-3)
No one (sensible) here is going to downvote you for asking why anyone would self driving cars.
You will earn plenty for preemptively bitching about downvotes. Just say your piece and don't worry about the little number next to your comment. If you're not a troll or a shill, generally your comments will be fine.
Upvote
24
(26
/
-2)
I'm not sure how Tesla is going to work with various municipal road, safety, fire, and EMS units without a policy team. Firing the policy team says "we are not in a business that requires complex regulatory oversight".
Upvote
23
(24
/
-1)
Also to do self driving properly at least with Waymo's approach it requires a significant amount of hardware. Add that having only a single seater restricts your potential market to single passengers with minimal baggage I think even purpose driven robo taxi's will end up looking like normal vehicles to maximize the customer base they can serve to cover the additional self driving hardware costs.
Upvote
12
(13
/
-1)
And that may be part of the training.
I've seen construction workers just wave their arms about and then get angry because the humans didn't understand their gestures.
There are standards for directing traffic, and if the workers can use those, the cars should be able to follow them.
At the moment, it's pretty random whether you'll get clear directions.
Upvote
11
(13
/
-2)
Speaking as someone who's driving career has just a decade or two remaining, I'd really like to have a car take over those tasks so I can continue to use a car. Not every trip is amenable to taxis, but having a car that can drive itself, or largely drive itself, will do wonders for the quality of life for older people.
Upvote
31
(32
/
-1)
I think that there are a few technical arguments in the article (Hint: read it slower). Besides, I don't agree with your choice of the term "political". It might apply if Trump, for instance, was behind those fabulous autocrashing devices, but it's Musk. And he is not involved in politics... Or is he?
Upvote
10
(12
/
-2)
Would you live your child alone with Elon?
A man who has sexually assaulted his workers.
A man who by all accounts is constantly on drugs.
A man who writes off killing animals needlessly.
A man who thinks killing police and firefighters is okay.
Not hating a guy who is a giant piece of shit in every way is just you highlighting that you are a womanizing racist.
Upvote
-15
(13
/
-28)
That's not the sort of training Tim was talking about in his article, though - he meant, Tesla training ES on their particular take on how their vehicles react, and their particular systems to handle them (apps, website, phone hotline, whatever).
By extension, Waymo have their own way, GM have theirs, etc. etc.
If you do a quick search, you'll find articles about first responders and Waymo and GM having dedicated 1800 hotlines... just ridiculous.
Fire truck arrives on scene... Oh, there's a Waymo nearby. Better call the number. And there's a GM... Jim, you call their number... etc. It's just totally ridiculous and unworkable for all but a very small handful of AVs across the entire city. (And, no, putting it on the dispatcher is no better - they'll have to call ALL the numbers/use ALL the apps/whatever - because they have no way of knowing which AVs will be in the vicinity.)
Upvote
9
(10
/
-1)
From autoexpress.co.uk on the new law for 2026
"One of the key elements of the new law, will be a crackdown on manufacturers misrepresenting their cars as ‘full self-driving’. This has already become an issue for some – namely Tesla – with the American giant recently caught up in a number of messy court cases in the U.S, which found that there was “reasonable evidence” that Elon Musk and other company executives knew that some elements of Tesla’s semi-autonomous system was defective."
"One of the key elements of the new law, will be a crackdown on manufacturers misrepresenting their cars as ‘full self-driving’. This has already become an issue for some – namely Tesla – with the American giant recently caught up in a number of messy court cases in the U.S, which found that there was “reasonable evidence” that Elon Musk and other company executives knew that some elements of Tesla’s semi-autonomous system was defective."
Last edited:
Upvote
10
(11
/
-1)
If you ignore all the a posteriori reasons, sure.
You act as if all of his last terrible behavior should constantly be ignored and we should look at each and every event with fresh eyes as if nothing bad has ever happened before.
But you knew the words "a priori," so you're obviously smarter than the rest of us.
Upvote
17
(19
/
-2)
Should you care enough for this reply, it's worth going back and reading my original comment that Emon replied to and their reply to me. Notice I was remarking on the article and they did not. I stand by my comment and position. Taken in context, I don't believe I was the unreasonable one here.
Edit: I'd also note that cybertruck, a pretty stupid vehicle, is not the subject of the article.
Upvote
-6
(4
/
-10)
ColdWetDog
Ars Legatus Legionis
Christ, if I could have the advantage of radar / lidar / extended spectrum visible light for driving, I'd love it. Plenty of times that the Mark I eyeball doesn't work as well as desired. Especially as we age out. And my brain, as limited as it is, is much better than the hardware in any car.
For a while, anyway.
Upvote
21
(22
/
-1)
I can't speak for all Arsians, but I dislike Musk because he has shown himself time and time again to be a thin-skinned, drug-addled, narcissistic bully and a blowhard who believes himself to be much smarter than his actions suggest, and has espoused some truly abhorrent political views on top of it all. I grant him no benefit of the doubt because his pie-in-the-sky pronouncements regarding his various company's products are practically a meme at this point (with the exception of SpaceX), and because he shows complete and utter contempt for any evidence contrary to his beliefs.
Many of us here were probably Musk and Tesla supporters early on because we liked the idea of commercially viable electric vehicles, and Tesla helped popularize that. However, Musk's antics over the past 6-8 years have dissuaded me (and likely many others) of any admiration of the man that I once possessed. It will take some truly momentous (and visible!) success to even begin repairing his credibility in my eyes.
He needs to put up or shut up, but he won't because he's incapable of either.
Upvote
40
(44
/
-4)
It's an engineering problem more than a theoretical one.
I drive with two eyes. I don't have more sensors (well, ears, but you could add those to a car too).
So you can operate a car on cameras. But that doesn't mean it's easy, nor does it mean we know how, nor does it mean it will work well at any point in the near future.
200,000,000,000+ per year in the US https://www.businessofapps.com/data/taxi-app-market/
But I think Tesla is trying, not to replace taxis so much as to replace cars. (I don't think is a great plan)
Also: the same sort of capabilities that work on taxies would work on, say, delivery vehicles.
Upvote
0
(9
/
-9)
Pretty sure that's a new Model 3 based on the headlights. Any details on what version/build of FSD you tested or how you accessed it? I assume Tesla didn't interact with you.
Upvote
8
(8
/
0)
When it comes to full self driving programs, does that fall under the licensed driver purview though?
I'm not being contentious, I'm honestly asking. It just doesn't sit well with me that there's no regulatory oversight for these systems.
Upvote
5
(6
/
-1)
fellow human
Ars Praefectus
It could be though. If you could just tap a button on your phone and a car arrives in a few minutes - affordably and without having a deal with a human - I think a lot of people would just do that rather than own a car, or maybe have one car instead of two. That's less cars on the road. Turning personal vehicles into public transport has a lot of upside imo.
It could be huge when it's finally possible.
Upvote
11
(13
/
-2)
Elon will be up late doing the counting with his family. In all seriousness, who is doing the counting?
Upvote
3
(5
/
-2)
Upvote
13
(15
/
-2)
ColdWetDog
Ars Legatus Legionis
Well, the idea isn't exactly wrong. Just not possible. The ground vehicle infrastructure is sadly in need of an upgrade into the 21st Century and infrastructure to vehicle and vehicle to vehicle communication would obviously be part of that. But it ain't gonna happen. Even if you could pry the money out, just getting all 48 states (and DC and PR and the Republics of California and Texas) to even come up with talking points would take us well into mid century.
So we're likely to end up with somebody walking in front of an autonomous car with a red flag (or laser light show given that we have progressed in some sectors).
Upvote
15
(15
/
0)
"While no one was behind the wheel during my rides, Waymo has remote operators that sometimes provide guidance to its vehicles (Waymo declined to tell me whether—or how often—remote operators intervened during my rides)."
Since waymo declined to tell the journalist how often, or if at all, it intervened in the supposedly self driving trip, how can we possibly make any determination on how advanced they are? Are we supposed to believe their press releases?
I'm sorry but I just don't buy it. With such opaqueness behind their self driving I don't believe any proper assessment or comparison of their capabilities can be made. For all we know a remote operator may have been driving the whole time to ensure a stellar experience and subsequent reporting by the reporter.
It's all meaningless without proper transparency.
Since waymo declined to tell the journalist how often, or if at all, it intervened in the supposedly self driving trip, how can we possibly make any determination on how advanced they are? Are we supposed to believe their press releases?
I'm sorry but I just don't buy it. With such opaqueness behind their self driving I don't believe any proper assessment or comparison of their capabilities can be made. For all we know a remote operator may have been driving the whole time to ensure a stellar experience and subsequent reporting by the reporter.
It's all meaningless without proper transparency.
Upvote
2
(9
/
-7)
I would think that having the cameras or RFID sensors on the cart itself would work. But then do we need multiple scannable labels or RFID chips on the box or can to ensure 100% accuracy. If so then we are increasing costs.
We are having major problems with self checkouts. There is rampant "shoplifting" as in not scanning expensive items, undercounting the number of items, etc. The tradeoff between labor savings and revenue loss due to shoplifting is no longer there. Not to mention the reluctance of local authorities to prosecute shoplifting cases from self checkout lanes.
Sometimes you just need a human to do the labor and nothing else would do.
Upvote
4
(6
/
-2)
But attached to those eyes is a visual processing system that's been under continual refinement for several million years just in primates and hominids alone, and a capacity for reasoning and judgement that you exercise at least most of the time. You're not just relying on your visual input. If there was a human-caliber brain in a Model 3, that would be one thing, but there isn't, so this is an incredibly weird assertion.
Upvote
32
(33
/
-1)
This is really the Musk mode - give your engineers an impossible task and watch your dream come true through blood, sweat, and tears - and abuse - of others.
It's kind of an effective sociopathic way of getting big things done.
One problem, though. Some things, just some things, are quite insurmountable with office heroics alone.
Upvote
12
(14
/
-2)
ColdWetDog
Ars Legatus Legionis
I think most people would be surprised about how much continuing education first responders have to go through. New buildings, roads, technologies, chemicals, equipment, lessons learned, regulations.
It isn't just washing trucks and eating pizza.
Upvote
30
(30
/
0)
The "humans drive with just their eyes so AVs can too" argument does sound logical at first, but the more in depth you go, the more skeptical I am. Unfortunately it's in ways LIDAR doesn't necessarily solve.
In addition to your point about the quality of human sight, I actually wonder how much we use other senses. If you pass a car going "whump-whump-whump" from a flat tire, or "skreeeeeee!!" From dragging something metal underneath, you're likely to give it a little more space if you think a tire is about to blow or an axle is about to fall off.
An AV's cameras might see smoke bellowing from a truck's brakes, but would it know that is a potentially riskier vehicle?
Additionally, it's not a traditional sense but at least some humans will use memory while driving like "yikes, that red Tahoe is the one I passed 5 minutes ago that cannot stay in their lane. If I pass them again I'm leaving extra space and doing it quickly!"
Do any AVs "remember" other drivers and/or flag riskier drivers and behave differently around them, or do they just assume every other car on the road is playing by the same set of rules and with the same motivations?
Upvote
32
(33
/
-1)
I see quite a few accidents involving gates closing. I can guarantee you there are multiple number of those involving human drivers trying to beat the gate closing on them. Of course if the gate had sensors so it does not close on a car...
Upvote
12
(13
/
-1)
The real prize is over-the-road trucking. Long-haul trucking doesn’t need an agent to deal with customers riding the truck the whole time. To a lesser extent, there would be significant revenue from sales to transit agencies that can dispatch a vehicle remotely and only need to send an employee if the user will require assistance.
Upvote
9
(9
/
0)
Long, long ago, in a galaxy far, far away, I was one of several senior technologists who convinced our employer not to participate in the Autonomous Land Vehicle part of the then DARPA Grand Challenge. Our argument was simple: machine vision is very hard. Machine vision in an environment of active threats is beyond very hard. Devote your resources to something with a better chance of success.
There are lots of merry pranksters out there with various agendas that would be furthered by deliberately pushing one or more of these vehicles to an edge case and over. It's going to be fun for them, not so much for the rest of us.
There are lots of merry pranksters out there with various agendas that would be furthered by deliberately pushing one or more of these vehicles to an edge case and over. It's going to be fun for them, not so much for the rest of us.
Upvote
11
(14
/
-3)
ColdWetDog
Ars Legatus Legionis
It isn't ridiculous, just not anywhere near optimal. If you have a Waymo, for example, doing something (or not doing something) and you don't remember the video you had on it six months ago and it's not on your first responder app or flash card, having somebody to talk to is a useful, low tech and cheap way of getting that information.
Certainly not optimal, but doable.
Upvote
10
(10
/
0)
enlightened.doggo
Ars Scholae Palatinae
I work in a highly regulated environment where our product is essentially ML, and I down-voted it because being an engineer doesn't grant someone the privileged to make up their own facts. The post as a whole would not be all that terrible if read in an information vacuum. We don't live in an information vacuum though...
Upvote
25
(26
/
-1)
Too bad it'll take much more than playing chess to get a fully realized self-driving car. It took decades longer for a computer to beat a human at Go than chess, and Waymo can't even fully complete a game of chess.
Upvote
-8
(2
/
-10)
As a general rule of thumb, one should trust autonomous driving systems proportionally to the amount of liability the manufacturer or service provider accepts.
Upvote
25
(26
/
-1)
Doable now, when there are like 10 AVs on the road in the whole city... when there are more? When there are more than 3 or 4 marques with AVs on the road?
And these are emergency responders - do we want them futzing around with AV manufacturers, or dealing with the emergency (including saving peoples' lives)?
It's totally short-sighted and, frankly, reckless to not have a single standard in place. My 2¢, anyway.
Upvote
0
(4
/
-4)
Yeah, I live in Mountain View where we've been seeing the various Waymo vehicles driving around for over a decade now - I think 12 years?
The one thing I have appreciated is that Waymo has (appeared to, per the results) consistently erred on the side of safety. If they're going to make compromises in favor of something, I'm fine with it being "not accidentally killing people" when we've seen so many other less conservative approaches doing exactly that.
Upvote
24
(25
/
-1)
Snarky Robot
Ars Legatus Legionis
The thing…it isn’t. I don’t give a shit if Tesla or Wayno “wins”. I’m probably not going to live to see a day where the average car is level 5 autonomous, and I’m young enough that I plan to be around a while. The problem isn’t sports all, it’s that Tesla’s approach is less safe and I have people in my life I don’t want killed because a self-aggrandizing bigot thinks his super big brain can outperform everyone else who came before him in any subject he spends 2 seconds thinking about.
I want Tesla to be BETTER, not for them to “lose”.
Also, for the record, declaring “I’m an engineer” when talking about things outside of engineering has, for a fair number of people, has come to mean “I will speak confidently about things I have absolutely no training or knowledge in”. Like, it’s a running gag that when I argue with people who make grand pronouncements in fields I am well acquainted with, and they refuse to listen, I ask them if they’re engineers. I think I’m somewhere in the 90% range for actually being right.
If you’re an engineer in a related field, go hog wild and let people know. Using it otherwise…it can lead to some really bad assumptions about the person. Because like most professions, being an engineer doesn’t give you insight into things outside that field.
Upvote
40
(41
/
-1)
- Status