How Uber Wasted $2.5 Billion on Self-Driving Cars (theinformation.com) 84
After five years and an investment of around $2.5 billion, Uber's effort to build a self-driving car has produced this: a car that can't drive more than half a mile without encountering a problem [Editor's note: the link may be paywalled; alternative source]. From a report: "The car doesn't drive well" and "struggles with simple routes and simple maneuvers," said a manager in the unit, in a 1,500-word email sent three weeks ago to Uber CEO Dara Khosrowshahi, warning of the issues. The self-drivingâ"car unit "has simply failed to evolve and produce meaningful progress in so long that something has to be said before a disaster befalls us," said the manager in the email, which The Information has seen. The manager -- whose identity The Information confirmed -- reflects a common belief across Uber that the unit, known as the Advanced Technologies Group, is destined to lose the high-stakes race to its rivals, which have demonstrated a lot more headway, comparatively speaking.
The manager is one of a growing number of voices, both insiders and former employees, raising concerns about the unit's effectiveness and questioning Khosrowshahi's ability to hold anyone accountable for its failures. Thuan Pham, Uber's longtime chief technology officer, who quit earlier this year, is one of those critics. "Over the past two years I have periodically raised concerns with Dara on whether meaningful self-driving progress is being made at ATG and specifically urged him to ask specific questions in this area in order to assess this for himself," said Pham in an interview. "I just don't understand why, from all observable measures, the thing isn't making progress. How come there hasn't been accountability or transparency?"
The manager is one of a growing number of voices, both insiders and former employees, raising concerns about the unit's effectiveness and questioning Khosrowshahi's ability to hold anyone accountable for its failures. Thuan Pham, Uber's longtime chief technology officer, who quit earlier this year, is one of those critics. "Over the past two years I have periodically raised concerns with Dara on whether meaningful self-driving progress is being made at ATG and specifically urged him to ask specific questions in this area in order to assess this for himself," said Pham in an interview. "I just don't understand why, from all observable measures, the thing isn't making progress. How come there hasn't been accountability or transparency?"
The initiative accomplished its real mission... (Score:5, Informative)
...namely to pump up Uber stock at a time it was slumping.
Re:The initiative accomplished its real mission... (Score:5, Insightful)
I thought success of the self-driving cars was the only plan they had to ever reach profitability? That's more important than the stock price; the stock price is largely irrelevant and would not drive major company decisions.
If the self-driving cars aren't the plan... tell me again why they're in a price war with taxis? If they "win," as soon as they raise prices to the actual market level, the taxi companies start offering service again. There is very little capital expense or financial risk in a taxi startup; just the cars. The drivers get commission and so don't cost anything unless they're making money.
Re: (Score:1)
...the stock price is largely irrelevant and would not drive major company decisions.
You've clearly never worked at a company where the executives are all committing massive securities fraud. Sometimes "Plan A" is to actually secretly scuttle the ship.
Re:The initiative accomplished its real mission... (Score:5, Informative)
Years ago, people complained that CEOs and other executives were getting paid enormous salaries even when the company was losing money. So everyone changed to stock-based compensation.
Now, every CEO has one and only one concern: "What can I do to make the stock price go up so that I make more money".
Self-driving vehicles are the new hot thing, so everyone is hyping the fuck out of them. It's all complete and total bullshit. It will be decades before we have *reliable* self-driving vehicles. But promising them "soon" is good enough to make your stock price go up and make you very rich (see: Tesla).
Re: (Score:2)
Self driving taxis are one of the harder problems to solve, but there are easier ones that will be done in the next few years.
For example shuttles at airports. Fixed routes, highly controlled environment. Haulage too, with pre-planned routes and vehicles that can operate nearly 24/7 (charge while unloading).
Re: (Score:3)
Re: (Score:2)
There are robot excavator systems too. You upload a map of what you'd like the site to look like and they make it happen.
Re: (Score:3)
Re: (Score:2)
But why work with Uber for these much simpler cases. We have other Automakers who can deal with that case as well. Heck they can go with John Deer, they have self driving tractors for over a decade now.
Re: The initiative accomplished its real mission.. (Score:1)
Re: (Score:2)
Uber is just Uber, their entire company is one big hypetrain, their work on sef driving cars is not indicative of the overall technological progress.
You mean just like computers beating humans at go? (Score:2)
It's all complete and total bullshit. It will be decades before we have *reliable* self-driving vehicles.
Oh, so you mean a similar timeframe and difficulty of computers beating ... errrm ... squishing humans at Go? [youtube.com].
Ok, joke aside. Just in case you've missed it:
Robots are already better at driving than humans. [youtube.com] And that video is 5 years old with some predicitons already shown to be too conservative. Because of economic pressure.
The only problem with the tech is that it's simply not yet mass-market-ready.
But e [youtube.com]
Re: (Score:2)
It may simple be that "all" the companies are trying to bash their heads against the wall with cheap sensors instead of LIDAR, and the companies showing off LIDAR systems are actually putting their development efforts into the visual spectrum sensors because they're cheap.
It isn't really clear that there is any reason for it not to work well. So far the accidents are mostly cases where they have shitty sensors, had the good sensors turned off, or had unfinished software that was ignoring the good sensors.
Th
Re: (Score:2)
There is a mistaken idea, that a Company's Primary Goal is to give profit to the shareholders. Some people think it is an actual law on the books. That isn't necessarily accurate, but a statement from a Harvard MBA professor who was sick of CEO's dragging the companies to the grown, while they made a lot of money. However as you have stated, the CEO had became the major shareholder, so the efforts had done a 180 and so much attention is towards the shareholders, that the company still isn't running optim
Re: (Score:2)
That's not true at all, executive salaries didn't go down when people complained they were high. WTF are you smoking?
Maybe that was just when you personally started learning about stock compensation, and it was actually already like this in the past? And maybe salaries still went up?
Re: (Score:2)
Re: (Score:2)
...the stock price is largely irrelevant and would not drive major company decisions.
You've clearly never worked at a company where the executives are all committing massive securities fraud. Sometimes "Plan A" is to actually secretly scuttle the ship.
You've clearly never worked at a US company where the executives are all committing massive securities fraud, because as you point out, those people do their planning in secret and you wouldn't know about it. You'd just be working at a company, like everybody else.
Also, working at a place where a thing happened doesn't give you extra insight in the thing, compared to somebody who reads about the thing. Actually, you're more likely to have personal biases in that case. The abstract analysis is generally high
Re:The initiative accomplished its real mission... (Score:5, Insightful)
Currently, Uber pays drivers and drivers cover the expense of their time and vehicles. The "sharing economy" argument is that you are using a vehicle that you already have, so the marginal costs of maintenance, insurance, etc. is small. I don't know the detailed analysis, but I understand that some Uber drivers actually make money while many effectively operate at a loss. This is all with losses at Uber big enough to bankrupt most states.
Now, with autonomous vehicles, someone has to own them. That means you have to purchase them and they will be expensive, especially at first. Then, you have to charge them or put gas in the tank. Next, you have to maintain them. This includes regular maintenance as well as roadside maintenance for the inevitable flat tires, breakdowns, accidents, or whatever. This requires people on location to go to the vehicles with tools and do some work. These people don't work for free. Finally, you need insurance. Maybe fleet insurance has different rates than individuals get, but you still have to be covered for liabilities.
Once you add in all these costs its not clear that this is less than the current scheme. Plus this is a COMPLETELY different business than what Uber operates now. Maybe they could contract much of the out, but then they will have to pay enough so that the contracting companies can operate. Of course, all of this assumes that autonomous vehicles actually work and are ready in a reasonable timeframe.
It has always felt to me that the self-driving car pitch was just a way to buy time and convince investors to give money to Uber.
Re: (Score:2)
Some googling suggests that an Uber driver makes about $9/hour after expenses and Uber's cut. Say the robotaxi can work 20 hours a day (between charging and maintenance) earning $9/hour. I know, demand isn't even 24/7, take it as an average.
That's $58,400/year. Fleet prices on the cars, overheads, let's say 2 years to break even. It looks viable.
Re: (Score:1)
If this ridiculous idea even starts to be implemented, a lot of riders are going to be very unhappy about having to brush crumbs or who-knows-what-else off their seats and kick trash aside so there's a place for their feet. Not to mention half the riders like having someone there to talk to. The whole thing
Re: (Score:2)
Also it would be the case that in self driving uber owned cars, People will just be more abusive to them in general. Oh you gotta pee. will just do it on the back seat, who's gonna stop you. It is just a car owned by a big faceless corporation, they can deal with some damage.
Re: (Score:2)
I keep hearing about how autonomous vehicles will make Uber profitable. Maybe I'm dense, but I don't see how that works without raising prices.
Currently, Uber pays drivers and drivers cover the expense of their time and vehicles. The "sharing economy" argument is that you are using a vehicle that you already have, so the marginal costs of maintenance, insurance, etc. is small. I don't know the detailed analysis, but I understand that some Uber drivers actually make money while many effectively operate at a loss. This is all with losses at Uber big enough to bankrupt most states.
You can't get from from conclusions to reasons for those conclusions via not knowing the detailed analysis.
Really you should start with potential reasons or ideas, and spitball some general numbers, make a tentative conclusion, and then do a more detailed analysis of the part you expect to be critical. But if you don't have numbers at all, you should leave out the economic characterizations as well, unless the numbers are obvious. But you admit they're not.
Robots cost more at initial hire than humans, that'
Re: (Score:1)
tell me again why they're in a price war with taxis? If they "win," as soon as they raise prices to the actual market level, the taxi companies start offering service again. There is very little capital expense or financial risk in a taxi startup; just the cars.
I don't understand the pricing either, since for the customer it isn't about price, it's about
* Modern hailing, knowing if/when the car will actually show up
* Not having to step out in front of the car on the street to get it to stop for you (yes, I've been to NYC)
* Entering the destination digitally
* Ease of payment, without the driver throwing hysterics because he wants cash he can skim
Many trips are either a) occasional or b) business; in both cases a moderate difference in cost isn't a gating factor, es
Re: (Score:2)
The thing is, none of those are problems in many places. Most people live in a regular size city, not a megacity. Taxis are cleaner, more reliable, can take you to places you don't know the address of, and 100% take credit cards. If they refused to accept payment, they'd get fired. I mean, you've got a phone in your pocket; call the same number you called to order the taxi, and tell them the problem. The radio-phone in the car will ring momentarily.
Where I am it isn't any of that, people usually choose some
Oh, did I predict this utter faceplant too? (Score:1)
Yes, yes I did.
Re: (Score:2)
Admittedly the effort seems to have been in a downward spiral since they ran over a vagrant, oh and there was the lawsuit from Waymo over stolen technology and ultimately...
Uber builds nothing, you can expect no positive growth from a company that exists to kill off an industry by undercutting them with some technology and workers that cannot earn a living wage on 40 hours of work.
Not that I was really a fan of taxi companies, but certainly a lot less rapey that Uber has been
Re: (Score:1)
Interestingly enough, I also predicted that, of the several companies scrambling to claim the title of first "self-driving" car to market, Uber's would be the first one to kill a vagrant.
AI is not easy, Research is long term (Score:2, Interesting)
I'm not sure that the short/medium term expectations of returns lends itself to this sort of research.
*B-)
It's not just Uber (Score:2)
Indeed! AI Winter II may be upon us as the "lacks common sense" problem is plaguing many AI products and investors realize they over-invested in AI, as we'll probably only get incremental improvements for a good while until the next breakthrough. Self driving cars made rapid progress at first, but all the self driving companies are struggling to make it practical now. Mistakes get wide press.
Re: (Score:2)
As anyone familiar with software development. for many apps, the Program is nearly 80% complete after a few week. Then you spend the next 10 months working on the 20%
You got the proof of concept done, you know the product can work. Then you have to work out all the stupid details and outlier conditions that ignoring can cause major problems.
Re: (Score:1)
This isn't like an office application where you are building yet another tracking and workflow system, because after a couple of decades you've done several tracking and workflow systems.
Nobody's ever made a working auto-car for public roads before. Thus, it's really hard to estimate the distance between prototype and practical production. Could be 5 years or 500.
This really does seem strange (Score:4, Interesting)
For a while I was going to a self-driving car tech meetup, I attended one meeting at a Uber a few years ago (last year?) where they went into pretty great detail about the approach Uber was taking, and showed us some of the hardware they were using...
To me what they were doing did not sound that much different than what other companies that have seen success, were also doing. So I am also puzzled why they have not made nearly as much progress as we have seen elsewhere (not just Tesla, but other companies like Waymo and others).
Maybe the terrible accident they had put too many safeguards or oversight around the system to where they couldn't possibly make it work.
Re: (Score:2)
Well there is this:
Former Google exec Anthony Levandowski sentenced to 18 months for stealing self-driving car secrets [theverge.com]
Which is to say they stole the technology and got caught, so probably nothing but an empty shell and a few crappy demos right now
Re: (Score:3, Informative)
That shows the real problem at the core. Most people missed it, self driving vehicles is a hardware problem and not a software problem, that is the delusion. There is no way in software to solve the self driving vehicle problem, it is a problem in hardware. How do you get the vehicle driving system, to see the world in three dimensions, in motion.
Once you solve that hardware problem, the software problem is easy. The computer can not see in 3D and is trying to best guess 2D into 3D, lots of wasted processin
Re: (Score:2)
That shows the real problem at the core. Most people missed it, self driving vehicles is a hardware problem and not a software problem
From what I saw Uber actually had a pretty good hardware program going though. They had a number of advanced LIDAR rigs, were testing some other stuff as well.. they were not lacking for hardware to address their efforts.
I've done some self driving car dabbling and to me, I really think it is largely a software problem. Hardware only goes so far bu the secret sauce is reall
Re: (Score:2)
All of the advanced efforts support 4D. (Score:1)
Pretty sure all advanced self driving car solutions at this time, have strong 4D modeling in place (i.e. predictive powers as to where an object will be in the future). The reason Tesla has done so well is they can test that model against all production cars on the road today, and see what the predictive engine predicted vs. what actually happened... Tesla explained this all in a very complete presentation a few years ago you can find on YouTube (it's 3-4 hours long).
Tesla also built custom hardware for th
It is a software problem (Score:2)
Computers have been able to see in 3D using stereo vision since the 1970s.
Recognizing objects, and deciding what to do, that is a much harder problem.
And hard problems require good teams of clever people with effective management. Otherwise they fail.
Re:This really does seem strange (Score:5, Insightful)
Re: (Score:2)
Humans have one massive advantage: Our brains.
When you look at a scene your brain recognizes everything in it, understands things like the shape of kerbs, the typical behaviour of pedestrians, that a bollard that has been knocked over and squashed is still a bollard and marks something important. It knows that there is no need to slam on the brakes for a plastic bag being blown into the road.
The human brain is incredibly good at making sense of what it sees, building it into a mental 3D map of the environme
Re: (Score:1)
Re:This really does seem strange (Score:4, Interesting)
God no, recognising objects is only the second-order problem that still needs solving. Recognising context is the third-order one, and we're nowhere near beginning on that one.
Context in this case is e.g. recognising that a suburban street has an increased risk of children suddenly appearing from behind a parked car, and adjusting your driving based on that. Humans can spot the character of the dwellings as housing families, and therefore expect children in traffic. No AI is even close to that level of contextualisation yet (and the current brute-force methods in machine learning may even be the wrong way to go about getting there).
Lots more advanced than you seem to think (Score:1)
Recognising context is the third-order one, and we're nowhere near beginning on that one.
All of the major efforts do consider context already, that's inherent in the training.
Context in this case is e.g. recognising that a suburban street has an increased risk of children suddenly appearing from behind a parked car,
Self driving cars today already slow down in situations like that since they know there's a greater possibility of issues occurring in front of the drive path, that they have lower visibility to
Re: (Score:2)
Object recognition is a huge issue that needs to be solved. When driving it becomes a hard real-time problem as well.
But note that humans perceive in 3D with or without binocular vision. In addition to the small parallax provided by having 2 eyes, we also use focal distance and inference, especially for moving objects.
Binocular vision really only contributes at close range.
You can test this easily enough. Close one eye. Can you still tell what's a picture vs what is an actual object in the room? Look at a
Re: (Score:2)
So utterly backward. Even in a perfect CG world with perfect perception the thing holding back self driving isn't the hardware it's the software on deciding what to do.
turd (Score:5, Informative)
read it yourself in their 2019 DOY financials:
https://s23.q4cdn.com/40796975... [q4cdn.com]
There's no way this ship can right itself. It's a bad idea, a crappy business model, it's a company that's horrible for the environment, adds to green house gases, contributes to traffic congestion, side steps local laws-licenses and fees, parasites off the lowest worker in the chain who need the money the most - the drivers - giving them subminimum wages without any benefits while the CEO sucks off $45M per year. Going somewhere you're not planning on going, taking people you don't know - for money - is NOT a ride share, it's a taxi.period.
I can lose billions too if you pay me $45M per year. Hell, I do it for 1/2 that... what a joke. investors are chumps.
Re: (Score:2)
Seriously. Where can I find investors like this willing to make a no-questions-asked multi-billion-dollar investment in my shitty business plan? At least my product would be technically feasible.
Re: (Score:1)
Re: (Score:2)
VC works on the principle that 9/10, what you're investing in will fail, but 1/10 you make an extremely high return. The downside for a VC is limited—whatever they put in. Depending on what the terms of the funding are, the upside is enormous. In the case of Uber, if they'd made good on what they were promising (no more cabs, dominance in ride-hailing), the return would have been insane.
Re: (Score:2)
is NOT a ride share, it's a taxi.period.
Seems to me how Uber cornered the market and put taxis in same category as kodak film and magnetic tape. Unless in a city like NYC with taxis in high density and all you need is a loud voice. But for many other places, an app on a phone is all you need. And don't need cash (I heard it was not until recent years cabs accepted debit cards). I've never used uber or lyft myself, except when I went to SF with some friends where we parked at a garage, she calls up for a Lyft ride. 30 seconds later someone is ther
Re: (Score:2)
Here's the one good idea they had: surge pricing.
Prices go up when things are busy, they go down when things are quiet. Drivers make more when things are busy, they make less when things are quiet. If the prices go up, drivers get themselves out onto the roads to capture some of the money, and they go home when demand is low and they can't make as much. This is a great idea on paper!
The problem is that people don't like variable pricing, and they feel scammed when the same trip costs different things at dif
Do we need more time? (Score:2)
From horrible culture, to C-level scandals and questionable behaviors, to lack of profitability.
Maybe it's time we admit that Uber is just not a good company. Even if you've made some money trading their stock.
I'm betting on Tesla (Score:2, Interesting)
Over the past few years I've seen a lot of skepticism about Tesla's approach. In particular people questioned whether a self-driving car really needs LIDAR.
Now we find out that the Uber car barely works, while meanwhile Tesla's self-driving software is good enough that some people are playing video games on their phone instead of watching the road. And they are getting tickets (as they should) but we aren't reading about horrible fiery crashes of death, because Tesla's self-driving more or less works alre
Re:I'm betting on Tesla (Score:5, Informative)
As well you should. If they get to "self driving", it will be in sunny daytime California conditions, with good lane markings. Let me know when they can self drive during a rural northern midwestern winter snowstorm, so I can stay off the roads they're driving on.
Re: (Score:2)
Re: (Score:2)
"Over the past few years I've seen a lot of skepticism about Tesla's approach." As well you should. If they get to "self driving", it will be in sunny daytime California conditions, with good lane markings. Let me know when they can self drive during a rural northern midwestern winter snowstorm, so I can stay off the roads they're driving on.
Initially it will certainly be restricted to sunny conditions. But the point of using a neural network is that eventually it does not need to see the lane markings [youtu.be] to know where the road is, just like a human driver does need them.
Re: (Score:1)
Re: (Score:3)
I think their sensors are inadequate for urban driving when there are one way streets and cars parked on the roadside. Depending on the right side B-pillar camera to see front cross traffic is dangerous especially when that traffic is approaching from the right (ie, when you're crossing a one way street). To see the same thing that a human driver (seated on the left seat) would see .. the car would have to move forward into traffic at least two extra feet.
Re: (Score:2)
To see the same thing that a human driver (seated on the left seat) would see .. the car would have to move forward into traffic at least two extra feet.
So you're saying that the car cannot have sensors on the front bumper, but a human can stick their head out through the front windscreen 7 feet or more like a goddamn giraffe? Are you high?
Re: (Score:2)
Huh? My point is that the Tesla NEEDS to have side-facing cameras on the front bumper. Instead of just on the B-PILLAR like it does now. What the hell, haven't you even seen a Tesla before -- there are no cameras on the front bumper. Go look up the camera locations. Also, the camera on the front fenders are rearward looking -- not side-facing.
Re: (Score:2)
So your eyes are on the front bumper? Isn't that painful?
Re: (Score:2)
Re: (Score:3)
Tesla is working on software to take two video camera views and work out a 3D view. They are calling it pseudo-LIDAR [cleantechnica.com].
What is it about Silicon Valley companies that they like to pretend to invent what's already out there?
What Tesla and the rest of the automotive computer vision scene calls pseudo-LIDAR, the rest of the world calls disparity mapping [wikipedia.org], and we've had it for decades.
I suspect Tesla's great strides in the 3D vision arena is due to the fact that patents on all the best 3D vision algorithms started expiring recently, and so those algorithms can be used without license fees.
Re: (Score:3)
Tesla don't appear to be solving the same problem. They certainly aren't providing any evidence that they can drive on anything other than highways.
Re: (Score:2)
Re: (Score:2)
So some highways and parking lots. (but apart from that, what have the romans ever done for us).
And how many accidents have occurred while these features are enabled?
I would like to see each vehicle evaluated by the same criteria, based on the same miles driven. Though at least California is forcing companies to provide some data [theverge.com]. Tesla isn't really participating.
Re: (Score:2)
Re: (Score:2)
Calling using two cameras psuedo-LIDAR is essentially fraudulent. It flatly does not do what LIDAR does. Cameras are simply vulnerable to certain failures that LIDAR is not. They're trying to construct depth data from visual information which is hard, and impossible when just one of two cameras is blind for whatever reason. LIDAR is inherently a depth sensor and that is why it is needed. If you don't have the complexity of the part of the brain responsible for processing vision (much of it in fact), you can
Uber’s business case at - risk (Score:3)
More than a taxi company Uber claimed self-driving ride hailing was its ONLY road to profitability. Ride share was its interim business model until the driverless car service launched.
Re: (Score:2)
Re: (Score:3)
Which is kind of mind-boggling, right? Uber have actually been totally upfront with this, and yet governments around the world eventually caved in and said ok, go ahead and decimate the existing taxi industry and other ride share platform competitors, it's cool.
It's not wasted (Score:2)
Uber's entire game is to try and get to self driving cars before that happens. If they do they win. If they don't they lose.
The stakes are pretty high, but so are the rewards. So the investors a
With hindsight ... (Score:2)
when does this become officially anti-competitive? (Score:3)
It seems like we're taking companies to task all over the world for being "anti-competitive" but not these chumps? From the outset Uber was intent on losing money and has never had a viable path to being profitable, and yet has been allowed and even encouraged to disrupt across the planet. They've thrown a ton of money at destroying the existing industries, no doubt killing off plenty of worthy opponents on the way with the sheer weight of investors. Now we're left with a smattering of gig economy platforms that put pressure on taxi operators, undermine public transport, cut corners within their own ranks, squeeze rights and earning power from workers and yet still bleed cash. The only winners will be those who were lucky enough to put money in before the stock was pumped and pull out before it dies.
Uber tends to put itself on the other side of anti-competitive lawsuits, particularly against the taxi industry. It's time to call them on their bullshit.
It's investors money wasted, not Uber (Score:4, Interesting)
From what I understand the company never made a profit (unless Hollywood accounting to avoid taxes), one of you here on slashdot said "they burn through cash faster than grandma at a one arm bandit." And the drivers at best can clear about $2/hr.
And from my files (I like to save notable posts from slashdot):
Re:Amazon's Self-Reinforcing Decline in Hires (Score:5, Insightful)
by Chris Johnson (580)
Sure, a bit. Uber's the same thing. It's designed to make maximum use of crazy people and force the others to live up to that standard or be fired.
I'll define 'crazy Uber people' not as 'danger to customers', but 'people who are bringing more value in terms of vehicle, skill and desire to please, than they are getting back in pay and benefits'. So the crazy Uber person is the one who keeps buying a new Lexus or whatever, vacuums their car three times a day and busts their ass to outperform all the other Uber drivers, so they can continue to win out over anybody else seeking to be a driver.
The key factor is that they are giving more than they get back, in the belief that they're cornering some kind of market or buying in to something important.
If you make a business that relies on people like this, you can demolish anybody else because you've worked out how to get voluntary unpaid labor, like the Amazon exec who was said to use her own money to hire subcontractors to do more. As long as there are people who are willing to do that, the market breaks and Amazon/Uber get to do what Wal-Mart did in small towns, break the back of other market participants so they can't break even or continue.
Another way to be a crazy Uber person is to put more depreciation and wear and tear on your car than you can afford to repair (or replace). It's easy to be crazy in these ways. It's externalities which are easy to overlook. These Amazon/Uber business models are designed to leverage that kind of crazy as hard as possible, and kick out everybody who's not willing to lose (one way or another) on the deal. Psychology is useful in getting people to buy into this stuff.
As they say, a cult.
Re: (Score:2)
So is anyone surprised about "concerns about the u
Re: (Score:2)
The saddest thing of all is that after Uber implodes they'll all get hired to run other companies into the ground.
Headline is bullshit (Score:2)
You do research to find out things. The money is _not_ wasted even if you have negative results. The idea that research must have positive results to be a "success" comes from people that have no understanding of what research is.
Literally billions, and it's a piece of garbage (Score:2)
AI (Score:2)
Ah, so you mean that the AI plateaued after a certain point and no longer provided significant gains without untraining / retraining it via a thousand times more effort than it took to train it to make a simple change, because the initial assumptions formulated in its first training stages have to be made minority cases in order to change its overall behaviour?
And that just throwing more data at it didn't make it any better? And that you can't tweak it because it's just a black-box of self-arrived-at arbit
If you've been following their effort (Score:2)
I'm pretty sure Uber was the company that almost got in serious trouble for hiring a Google engineer who proceeded to try to steal Google's self-driving tech. Also the same company that had an autonomous vehicle run someone over. So no wonder this department is in dire straits. Google's currently got the best self-driving system and even they don't feel it is quite ready yet.
Balloon to the moon (Score:2)
"You failed to produce a balloon that could go to the moon. Off with your head!"
"before a disaster befalls us" (Score:1)