AI = Actual International-workers
AI = Asians Inside
this is actually a great advert for waymo ai. taking a safety first approach better than humans can and referring to humans when in a situation in which it has no safe option
AI =Actually Indians
I’m also AI
Actually Introverted
The Mechanical Turk strikes again.
Waymo knowing when it is stumped is actually a pretty good thing. Better than just running over cats & small children.
Curious what the law is with regard to someone in the Philipines driving a car on US roads without a US driver’s license.
It’s a mega corpo. Laws don’t apply to them silly you.
Laws are for poors.
With like 200 ping
Maybe the same as tourists driving in the US without US licenses.
You will ignore the man behind the curtain!
The answer is of course proper regulation. Too bad that is ceasing to exist.
For anyone that is curious, Waymo actually is capable of remote moving the vehicles despite what they say. They do their best not to admit it’s possible, but it’s right in the CPUC filings as a footnote, and probably the only place they’ll ever admit it.
In very limited circumstances such as to facilitate movement of the AV out of a freeway lane onto an adjacent shoulder, if possible, our Event Response agents are able to remotely move the Waymo AV under strict parameters, including at a very low speed over a very short distance.
I’m not opposed or knocking that they can do this, but they are lying to or misleading people when they say it can’t be done.
Waymo really seems to be winning out over Tesla with the self-driving thing. I wonder how much of that is really just because Waymo cars have a remote human driving them in situations where a Tesla would just crap out
I’m not fully up to speed on Waymo and if they have ever released remote assistance/ miles details, but when Cruise went through that shit storm a year or two ago, it came out that that the cars were asking for help every few miles.
Cruise was essentially all smoke and mirrors.
Interesting. Stuff like this makes me suspicious of the current LLM hype. I know it’s not necessarily language models per se being used by these vehicles, but still. If we were really on the cusp of AGI then I’d expect us to have at least cracked autonomous driving by now.
Ya, I don’t buy the hype around AGI. Like a Waymo drove into a telephone pole because of something they had to fix in their code. I’m not doubting there’s AI involved, neural nets, machine learning, whatever, but this isn’t an AGI type level development. Nor do I think they need an AGI to do this.
I’m also not convinced this LLM stuff can ever lead to AGI either. I think it can do some pretty impressive things with some very real drawbacks/caveats and there is definitely room to keep improving them, but that the whole architecture is flawed if you want to make an AGI.
Nor do I think they need an AGI to do this.
Yeah I guess theres a lot of interesting stuff we can do with AI without necessarily achieving AGI. What about programming? Even if we don’t get AGI soon, do you still think LLMs will be snatching up a sizeable chunk of programming jobs?
So I’m developer, I do mobile apps, and I do use Claude/GPT.
I could be wrong, but I don’t foresee any imminent collapse of developer jobs, but it does have its uses. I think if anything it’ll be fewer lower end positions, but if you don’t hire and teach new devs, that’s going to have repercussions down the road.
I needed to make a webpage for example, and I’m not a webdev, and it helped me create a static landing webpage. I can tell that the webpage code is pretty shitty, but it does work for it’s purposes. This either replaced a significant amount of time learning how to do it, or replaced me hiring a contractor to do it. But I also am not really any better off at writing a webpage if I needed to make a 2nd one having used it, as I didn’t lean much in the process.
But setting it all up also did have me have to work on the infrastructure behind it. The AI was able to help guide me through that as well, but it did less of it. That I did learn, and would be able to leverage that for future work.
When it comes to my actual mobile work, I don’t like asking it do anything substantial as the quality is usually pretty low. I might ask it to build a skeleton of something that I can fill out, I’ll often ask it’s opinions on a small piece of code I wrote and look for a better way to write it, and in that case it has helped me learn new things. I’ll also talk to it about planning something out and getting some insights on the topic before I write any code.
It gives almost as many wrong/flawed answers as right answers if there’s even a tiny bit of complexity, so you need to know how to sift through the crap which you won’t know if you aren’t a developer. It will tell you APIs exist that don’t. It will recommend APIs that were deprecated years ago. The list goes on and on and on. This also happened while I was making the webpage, so my developer skills were still required to get to the end product I wanted.
I can’t see how it will replace a sizeable chunk of developers yet, but I think if used properly, it could enhance existing devs and lead to fewer hires needed.
When I hear things like 30% of Microsoft code is now written by AI, it makes sense why shit is breaking all the time and quality is going down. They’re forcing it to do what it can’t do yet.
Thank you for the detailed response. Most opinions on this topic are very much so based on vibes rather than real experience, so it’s interesting to hear an informed opinion from someone on the inside.
I hope to become a software developer one day too (it’s a slow process, because I’m teaching myself in my free time) so I sometimes worry if all the effort I’m putting in is even worth it if LLMs will be doing all the programming in a few years. Do you think that’s a concern? Will these tools continue to develop to that point or are they hitting a wall, like some people are saying?
And these foreign crowd workers know the local traffic rules? Maybe they even have regular drivers licenses?
I think the interventions here are more like: “that’s a trash can someone pushed onto the road - let me help you around it” rather than: “let me drive you all the way to your destination.”
It’s usually not the genuinely hard stuff that stumps AI drivers - it’s the really stupid, obvious things it simply never encountered in its training data before.
Saw this blog post recently about waymo’s sim setup for generating synthetic data and they really do seem to be generating pretty much everything in existence. The level of generalization of the model they seem to be using is either shockingly low or they abort immediately at the earliest sign of high perplexity.
I’m guessing it’s the latter, they need to keep accidents to a minimum if they’re ever going to get broad legislation to legalise them.
Every single accident is analysed to death by the media and onlookers alike, with a large group of people wanting it to fail.
This is a prime example, we’ve known about the human intervention for a while now but period people seem surprised that those people are in another country.
Feels like the robot hoovers when they encounter an unexpected poo.
This used to be my job. They’re not controlling the cars. They’re basically completing real-time CAPTCHAs, telling the car whether the cameras see a stop sign, a bicycle, temporary barriers, etc. If the car can’t identify an object that could possibly cross its path, it pulls over and stops until an operator can do a sanity-check on whatever the car’s confused by. They only need to be able to identify objects on the road, not know the rules of the road.
Can you imagine the lawsuits?
No. I am not from there. Feel free to explain what is possible.
In my country we have a law that requires such remote operators to have a license that is valid here.
(Sadly, we do not require them to reside here)
deleted by creator
This would have actually been a great thing to not only acknowledge but promote if they weren’t so caught up in their own hype.
Not that I will ever get into one of those death traps but if you tell the average consumer that any failures in autonomy immediately engage a tele-operator “to keep you moving on your way” they would probably feel better about riding.
I’ve done tele-driving before and it’s remarkably good, even if latency is a concern.
It’s the facade of it all, the need to seem to live up to the hype. It’s going to get more people killed.
any failures in autonomy immediately engage a tele-operator
One of the problems is that these “failures in autonomy” could include a failure to engage a tele-operator when one is needed.
Nothing could make me feel better about my vehicle being operated remotely by someone in another country. Granted, nothing could make me feel better about my vehicle being operated by a computer either. I’ll drive my damn self, thank you.
I don’t know I trust these over most of the drivers. Hell I took a lyft last night and on the freeway we were pretty scared with the way the driver was going.
I work near downtown Austin, where both Waymo and Robotaxi operate.
Waymo cars are some of the best drivers on the road because they actuallyt ested their product, use multiple Lidar sensors instead of just cameras, and have remote driver backups for unusual situations.
Teslas drive like maniacs and will end a ride and tell the driver to get out in the middle of a lane.
Automation has always been about de-skilling to cheaper, more abuse-able labour, and not about actually eliminating work. This goes all the way back to the broad looms and the luddites. There were still loom workers in the new factories - its just that they were children who could be worked to death for pennies.
Well, automation under our current system, yes.
Moravy also argued that to stop anybody from taking control of vehicles, the company “actively participates in hacking events
Read this slowly:
Here they admit that their vehicles can be hacked and then remotely cotrolled.
They’re running red team hacking scenarios, an extremely standard, common, and good practice in the cybersecurity industry. Any device, especially one connected to the internet, is at risk of being hacked - it would be naive to assume otherwise, so they’re hiring professionals to penetrate their security before someone else does. This is actually a sign they’re taking security seriously.
Also, from the article: “they do not remotely drive the vehicles”.
Also, from the article: “they do not remotely drive the vehicles”.
You may quote and repeat this as much as you like… ;-)
Do you actually have any evidence Waymo staff can remotely drive their vehicles? Or are you just tilting at windmills? I don’t really appreciate the insinuation that I am some rube by someone evidently unaware of basic cybersecurity concepts.
There are real problems with this arrangement that should be focused on rather than vague speculations - i.e. the exploitation of developing nations by the machine learning/tech industry.
(not op) Right here. It’s the only place they’ve ever admitted its possible.
In very limited circumstances such as to facilitate movement of the AV out of a freeway lane onto an adjacent shoulder, if possible, our Event Response agents are able to remotely move the Waymo AV under strict parameters, including at a very low speed over a very short distance.
Hmm interesting, thanks for the link!
Do you actually have any evidence
I have expressed my belief, or my doubt, however you want to look at it.
someone evidently unaware of basic cybersecurity concepts.
I can assure you that is not the case. I work in IT, all my life, much longer than you, and I know all of it’s basic concepts.
Think: what would happen when such a car gets stuck and the remote operator can’t achieve anything with “giving directions”? He needs some stronger action. Maybe he needs to “escalate” to some “senior”. What would that person do?
There is the possibility of remote steering, and I think they would use it, 10 out of 10 times, instead of telling their passengers that they give up now and everybody must leave the car.
For future reference, here’s your proof its possible to be remotely moved, which means a hacker could exploit it.
In very limited circumstances such as to facilitate movement of the AV out of a freeway lane onto an adjacent shoulder, if possible, our Event Response agents are able to remotely move the Waymo AV under strict parameters, including at a very low speed over a very short distance.
“I work in IT, all my life, much longer than you, and I know all its basic concepts.”
And anyone who’s spent their life working in IT would laugh you out of the room for that sentence. Lost all credibility with that BS.
No. They just end the ride and send somebody from the local depot to drive the car back to the garage.
Source: I was on Waymo’s Fleet Response team for a year doing literally this job that is now outsourced overseas. While the tech exists for full remote steering, NHTSA regulations disallow it, and that’s one of the few agencies that Google actually has to abide by if they want to drive their cars on public roads.
Source: I was on Waymo’s Fleet Response team for a year doing literally this job
Good to hear. Thanks for sharing this.
But still, if I were some higher manager there, then I would probably think a little different than you honest people:
that is now outsourced overseas.
- One of the differences is that these operator people come a lot cheaper now.
NHTSA regulations disallow it
- Another difference is that they won’t ever tell any American what they actually do at their job, because they are on the other side of the globe, where it makes no difference at all if they can spell this 5 letter abbreviation or not.
“Your kid is safe at school, because we insist all teachers use condoms”
I still have no idea how these are legally able to operate on public roads. Shit seems wild to me. Wouldn’t last 5 seconds here in Chicago, for numerous reasons lol
Why won’t waymo work in Chicago specifically?
“Hey clanker, you drove in the wrong city…!”
