mikejb

This is from from November 2021, but I'm still highlighting it because it is just terrifying (Correct time, though the video later on also exhibits inabilities of the system): https://youtu.be/9wRRClg_aM8?t=113

I think it's not just that. When Tesla started to with their FSD journey, they had to determine what sensors they can add to the car. Lidars back then were way more expensive than they are now, and it wouldn't have been feasible to add them at that time.

They can't add them now to new vehicles because they promised the vehicles back then are only a software update away from full autonomy [1]. Building on Lidar now would mean developing on 2 heavily differentiating stacks. Going back on the promise of old Tesla's being "FSD capable" would introduce a huge liability.

Long story short, Tesla's stance on Lidar determined 8 years ago, without the option to revise the decision with future developments.

[1] Note this has turned into "we'll only have to replace the computer in the car", which is still doable, contrary to adding sensors to the existing vehicle.

Another country that comes to mind is India - but I've never been enough of a Bollywood fan to be able to make any assessment on this topic.

But it's a very effective tool to shape public opinion. I'm fairly certain without the overwhelmingly heroic portrait of (members of) the military, the public would be more inclined to question the huge investments that go into warfare. Particularly the display of highly achieved individuals in the military leads to both public respect of military members, and enlistings to service.

As much as it would be a solution to Tesla's problem, it would impact Tesla's overall plan and strategy. Automatically disengaging when emergency vehicles are detected is a huge barrier in their selling strategy ("It can drive itself, it's just not allowed yet to for legal reasons, but soon").

> I'm not seeing this with the current situation

It's what Waymo's is currently doing. So far, you only can get automated Taxis (in some regions) - you cannot buy an automated vehicle yet.

As I see we do agree that they will exist - and we also agree that they will never have 100% worldwide adoption. That would be a ludicrous assumption. Someone not buying a car now will not buy a self-driving car either. Someone never taking a cab will not take a driverless cab. I don't know how that was assumed or where that was ever claimed.

I think this also translates to self-driving cars. I actually think automated cars have an edge there, because of 2 reasons:

1. Improved visibility: A human driver sits in the driver seat, and looks at the outside world from inside the car - partly through mirrors, partly with the view obscured. A self-driving car (at last current potential vehicles) has a view without being obstructed by the car's frame.

2. No limitation on focus: As a human driver, I can look to the front, or the back, or the sides, but I can't look everywhere all the time. I can do my best to maintain an overview of my surroundings, but need to update this constantly by changing where I'm looking - if I want to see if there's someone behind me, I (for a moment) have to stop looking where I'm going. Automated vehicles don't need to stop looking to the front to see to their back or the side. They can see a car 200 feet away coming from the right without having to miss the pedestrian coming from the left.

I'm not saying that self-proclaimed self-driving cars do this already, but looking at perception capabilities, self-driving cars have the potential to avoid more accidents than humans.

I think you're confusing a Tesla with a self-driving car.

Ok, I think we're having 2 separate conversations:

a) What's the impact of self-driving cars at what point of distribution

and b) What's the distribution of self-driving cars going to be.

Regarding the impact: modern vehicles are getting better at minimizing the impact of accidents, and are starting to avoid accidents, mostly in the form of avoiding rear-ending another vehicle. But Self-driving cars are on another level there: They can recognize someone possibly running a red light, and act accordingly (not entering the intersection, evacuating the intersection quickly, etc.). They have sensors to continuously monitor their surroundings with a focus a human can't have, and modern vehicles don't bring the sensors for it because of the actions a human-driven vehicle can take are highly limited (Maybe a warning-beep, tightening seatbelts and prepping to fire airbags - but nothing in terms of avoidance. And you need a fairly fancy new vehicle for that). Additionally, also drivers of expensive / new vehicles crash. They have phones like anyone else, drink like anyone else, get tired or distracted like anyone else.

Regarding the distribution of self-driving cars: As I said, it's not feasible to assume that by the time you can buy or rent a self-driving car in the bay area, you'll also be able to buy or rent a self-driving car in Mogadishu. Additionally, ownership is one aspect of automated vehicles. Companies also aim for a service-type of business, where you basically take a cab - just without a driver (who again can get distracted, tired, etc.)

Now, if you say "we're a century away from having 50% self-driving cars in Somalia" - I wouldn't disagree. But the comment that started this discussion (and to which you said you have more or less the same opinion) questioned their present and future existence:

> Self-driving cars do not (and will not, if you ask me) exist

Maybe we're disagreeing on a misunderstanding?

Auotmated drivers don't make the human driver safer - they (should) make traffic safer by reducing the number of human drivers that would cause accidents. I don't think we're there yet, but I also don't think this is centuries away, based on the progress made in recent years. I also don't think this will be instantly world-wide: Distribution of automated vehicles will be just like the distribution of any other commodity. In some parts of this world it's infeasible to have a car for a significant portion of the population - it'd be ridiculous to assume those parts of the world will have self-driving cars by the time they're available in the Bay Area.

It's also inevitable that human and automated drivers will share infrastructure. That's where the development is going, it's what will happen. There won't be a switch like "from tomorrow on, only automated vehicles can drive in San Francisco".

I agree with the mass transportation system investment, particularly in the US. But if you argue against financial feasibility of self-driving cars in regions, you have to apply the same scrutiny to public transport investments.

I think Lvl3 is one of the most dangerous levels a car can be at, based on the experience Waymo (back then Chauffeur) made: Humans are even worse at monitoring than they are at driving. It's incredibly boring. Their test-drivers (who got extensive training for what to do and what not to do) started to do anything but pay attention, including taking a nap, before they pulled the plug on the incremental approach.

And we see this partially with Tesla now. Some drivers start to trust the system so much that they start to do anything but pay attention - partially even bypassing the pesky systems that remind them that they are required to take over at any moment.

To be fair: Self-driving cars that can drive safer than some drivers already exist. Simply due to the nature how unsafe some humans drive. (Some of these drivers are elderly, some of these drivers are racers who think everyone slower than them drives like a grandma).

I'd be interested though why you think self-driving cars will not exist. Care to elaborate?

I have my doubts about V2V and I2V communications. They'd simplify the task for automated systems, but come with a rat tail of problems.

I2V is incredibly hard to scale, and is a duplicated implementation of a pre-existing system (road markings & signs). Any system with duplicated implementations will grow inconsistencies. In those situations, the automated driver will be driving based on a differently perceived environment than the human driver - and it'll be hard to know. The automated system would also need to read lane markings & signs to recognize these inconsistencies and act accordingly, at which point the I2V communication is just a crutch to help the automated vehicle, reducing it's utility and hence the cost-value ratio (which is already insanely high).

V2V communication is unreliable by the fact that older vehicles won't support it, so automated vehicles won't be able to communicate with some (initially: most) cars. They'll need to be able to drive safely without communicating with nearby vehicles, so V2V is again just a "bonus" system a vehicle must not rely on.

Automated drivers will be safer than human drivers with time. It's an incredibly hard problem to solve, but it will be solved. Technology is improving continuously, and different companies are trying different approaches to solve the problem - which makes me optimistic in terms of automated drivers. Becoming safer than human drivers is mostly a matter of being able to drive with an automated system based on infrastructure optimized for humans. Sources of accidents will shift: From driver impairment, distraction and ignorance of rules, to misinterpretation of the environment. Additionally, and I think this is the key: If an accident happens, at best the involved human drivers improve their reaction to the specific situation. Automated drivers can share this new knowledge across the entire fleet.

That particular attack had been fixed, but I remember people disabling those warnings with oranges: https://www.youtube.com/watch?v=TYZrehVQouc

It's an arms race between Tesla being legally on the safe side, and drivers using the car as Tesla advertises it.

> you always have plenty of warning before you need to take over

This only has to be false once.

I'd also like to highlight that sometimes AP has frightening regressions. One case was labeled "barrier lust", where - on a divider - AP would follow the wrong white line, and steer the car towards a barrier. See here: https://www.reddit.com/r/teslamotors/comments/8a0jfh/autopil...

It was fixed, after Walter Huang died through such a case.

Until the behavior returned some months later: https://www.reddit.com/r/teslamotors/comments/b36x27/its_bac...

So please be careful. Regardless of the value you put on your own life, car accidents impact more than just yourself.

Right, because who would digitalize the entire world and take pictures of every house and store front - that'd be insane!

This is a curious point: How much data does Tesla actually collect? I can imagine users not being too happy if their car clogs the upload for hours every time you park it in your garage?

To be fair, Waymo (back then Chauffeur) have solid experience with Tesla's approach. They started with the same strategy of gradual improvements. They found out (and shard) the same issues Tesla is running into in the past years: Users don't pay attention when monitoring a system doing a boring task - even when failures carry a high price. Waymo switched the approach from a gradual L2/3/4 improvements straight to developing an L4 system; Tesla is locked in to the gradual move due to it being part of their business model, and are now stuck with trying to make self-driving work with unqualified test drivers (i.e. customers) and outdated technical decisions (not using Lidar because they were too expensive back then).

So Waymo does have the experience to call out Tesla here.

Whilst Lidar and cameras may have similar wavelengths, they have very different information outputs. Cameras give you a 2D colored image from which you essentially guesstimate the distance of objects. Lidar gives you a 2D depth map with much higher confidence in the data that comes out of the stream.

Elon has a high incentive to talk down Lidar in cars: He/Tesla placed the bet initially on cameras because Lidar is/was too expensive, and promised these cars will be able to drive themselves ("regulatory approval pending"). Retrofitting Lidar is not an option (price, sensor&cable placement), so a lot of his (and Tesla's) promises ride on cameras working out and Lidar not being necessary. A lot of Tesla's (Car's and Car company's) value rely on cameras working out and Lidar not being necessary.

Your 'experience' explains a lot about your opinion.

Have a good day, stay safe.

So you say you suspect they have an ongoing investigation, and that would be "circumstances where every half competent organization would decline comment"

What substantiates this conclusion? They can have an investigation ongoing, and share the cause for said investigation. In the statement they explicitly establish that this doesn't imply guilt of the account owner.

Why are you so triggered by this clarification?

> And even if the ethical obligations were identical, Gebru’s violation toward Google wouldn’t excuse Google’s toward Mitchell.

What about Mitchell's violation towards Google?

Just to get this straight:

The researcher who tweeted out the name of an employee who's email has been blocked, and throwing out theories about crackdowns and firings - that's all good.

Explaining the email account has been blocked due to mass-leaking documents - that's beyond excusable?

Sure. When facts contradict your opinion, you shouldn't hate the facts.

> The normal response (for a variety of good reasons, including legal and ethical ones) from a company on a matter like this when asked by the press when they haven’t completed an internal investigation would be to decline comment on personnel matters. Google’s behavior is grossly unethical here irrespective of who reached out to the media first.

To give a counter-argument to this: If party A makes a claim or accusation, and party B just says "no comment for now", it'll be near-impossible to reduce the (potentially unjustified) fallout of the premature verdict made. Public opinions matter, and stories told one-sided should not be desired by anyone interested in an objective discourse.

The researcher who's departure sparked a lot of controversy tweeted the name of the employee in question, with some opinion attached. Axios reached out to Google who then made the statement saying security systems detected mass-forwarding of internal documents. They didn't name the employee, but prefaced the explanation with "in this instance".

Not every leak is a whistleblow. It depends on the content of the files. That verdict still needs to be made (depending on the content of the leaked files).

The (1) case is a bit inaccurate/misleading. From what I can gather from the article:

- Gebru tweeted the name of the employee [1]

- Axios then reached out to Google, who then made the following statement:

> Our security systems automatically lock an employee’s corporate account when they detect that the account is at risk of compromise due to credential problems or when an automated rule involving the handling of sensitive data has been triggered. In this instance, yesterday our systems detected that an account had exfiltrated thousands of files and shared them with multiple external accounts. We explained this to the employee earlier today.

Context matters.

[1] https://twitter.com/timnitGebru/status/1351698317550432256

I couldn't find the direct quote, so they may have adjusted the sentence you're referring to. It's clearer now:

> If you're inactive in one or more of these services for two years (24 months), Google may delete the content in the product(s) in which you're inactive.

Lidar-based systems are pretty good at detecting stationary objects. Humans on the other hand crash into traffic jams pretty regularly.

So in that situation I'd prefer that to be an automated truck.

I'd say the breakthrough term was not the ideal word choice - it's more that technological advances passed a threshold:

For VR, a lot was advances in (affordable) processing power (latency causes significant issues in VR), and screen (manufacturing) technology (i.e. pricing improvements).

For self-driving, it's again processing power (with the focus on the ratio between computation power and power consumption/size), but also the advances in technologies like Lidar, as well as advances in theoretical research (and better understanding) of machine learning.

I named these 2 companies because the parent referenced them in their argument. But I agree that they are the ones that are the ahead in their individual approaches, though I think Tesla is not really that near to autonomous driving. They have a lot of features in development that their test drivers (=consumers) are using, and that gives the impression that they got that already (Elon even claimed Tesla has FSD on highways in early 2019), but the step to "the driver doesn't always have to pay attention" is gigantic and far away - and on that scale, Tesla's lead to other manufacturers (who don't do that kind of marketing or testing on consumers) is relatively small.

They're still doing better than a lot of the wannabe-unicorns that pursue self-driving or Lidar though.