Posted on 08/14/2019 7:29:51 AM PDT by C19fan
Uber set two dubious quarterly records on Thursday as it reported its results: its largest-ever loss, exceeding $5 billion, and its slowest-ever revenue growth.
The double whammy immediately renewed questions about the prospects for the company, the worlds biggest ride-hailing business. Uber has been dogged by concerns about sluggish sales and whether it can make money, worries that were compounded by a disappointing initial public offering in May.
For the second quarter, Uber said it lost $5.2 billion, the largest loss since it began disclosing limited financial data in 2017. A majority of that about $3.9 billion was caused by stock-based compensation that Uber paid its employees after its I.P.O.
(Excerpt) Read more at nytimes.com ...
LOL!
Its not the programming... its the PROBLEM...
I am a software engineer, and I know many people working in the autonomous driving industry at a lot of different companies, and the reality is, ITS DECADES AWAY....
The list of “problems” involved in autonomous driving on an interstate, let alone the city, that folks have ABSOLUTELY NO IDEA HOW TO SOLVE, is endless.
Autonomous Vehicles are an 80-20/90-10 problem. Yes its come a long way in the last 25 years... but thats only been 80-90% of the problem... but that only represents about 10-20% of the actual problem, and the amount of effort to solve the last 10-20% of the problem is going to take 80-90% of the overall effort.
AI/Machine Learning isn’t up to the task. People think there is some sort of super tech in this and it will miraculously solve everything.. and while it has great potential, it is NOT a miracle. The machine simply comes up with a probability based on its training as to WHAT something is. It cannot and does not infer or logically solve anything. If the algorithm says the woman walking across the street with her bicycle isn’t something to worry about. or can’t really decide.. then boom... woman is run over, even though every other human controlled vehicle on the road avoids her.
If the computer says the semi crossing the highway at the top of the hill ahead of you is an overhead sign... then you get decapitated at 70 miles an hour as the brilliant computer decides it can drive right under it.
Simply driving a semi on the interstate is far simpler than driving city driving, no cross traffic, everyone traveling the same direction, only need to get on and get off and change lanes safely and not rear end anyone... and yet there isn’t a single certified autonomous truck on the road.. Why? BECAUSE THEY HAVEN’T AND CAN’T SOLVE THE SUBSET OF PROBLEMS THAT EXIST IN THIS MUCH SIMPLER USE CASE... Autonomous vehicles on city streets are death traps...They are NOT remotely safe, and its honestly a miracle so few people have been killed by or in them so far.
This tech is DECADES away...
Now, with that said. UBER’s other problem is, even once they have autonomous vehicles, their entire first mover advantage disappears almost instantly as their entire business is now a commodity....
Throwing money at UBER other than to flip it to a bigger fool short term, is a fools errand long term.
As I said, Field work.
How can you program anything effectively if you have not collected data on every possible scenario?
+. Good summary!
+. Good summary!
+. Good summary!
Well said. I agree 100%.
I agree with you completely.
I’m an old R&D guy.
I’ve heard about built-in fail-safes for autonomous cars that would allow the human pilot to take over in case things start looking bad.
I have a lot of doubts about that. For one thing, it would require a human who is not driving to pay as much attention to the road as someone who *is* driving. That isn’t going to happen. Second, it would assume that the person taking over in an emergency would have the driving skills to safely navigate the dangerous situation. Presumably, this person who has all this skill is someone who might have been taught to drive, but their experience is as a passive passenger in a car with no experience on how to drive in normal traffic, let alone when there is a situation.
I’m worried enough about the wisdom of introducing technology that will automatically engage the brakes when the car senses it is too close to an object. Won’t that encourage drivers to be less attentive, since they know their cars will stop automatically regardless of whether they are attentive or not? What happens when they are used to driving a car that automatically stops for them, and they drive a car without that feature. Oops, the guy ahead just slammed on his brakes, and they rear end him because they were only half paying attention to the road, expecting their car to handle such issues.
I see a lot of problems with self-driving cars. They would need to have human intelligence to assess and react in a split second, but they are just computers, and they only do what they are programed to do.
I am reminded of George Costanza running over the pigeon and the squirrel, and all the hell he paid for that decision.
$5 BILLION loss?
No problem.
They’ll make it up in volume.
Do I hit the dog? or dive into the other lane and kill myself?
A friend and I were talking about one of the first pocket calculators, the Bowmar Brain. We each had one of those decades ago. The Bowmar company is long out of business. So I see your point.
Slide rulers got us to the moon.
But it was pilots seat of the pants that landed the LEM
There in is the RUB.. you CAN’T collect data on EVERY scenario... its impossible.
The list of problems they can’t solve, even when they have data is HUGE... I don’t think you get what I am trying to tell you.
Take a simple speed limit sign, that displays different speeds for different car types for different times of day. Its a single sign conveying all sorts of information, and there is no standard way this information is conveyed.. A sign doing this in one place is completely different from a sign doing the same thing somewhere else... and lets not even get into the ones that change their values based on traffic and other conditions dynamically.
You as a human can see that sign and know quickly what it is conveying to you... ML/AI doesn’t they HAVE NO IDEA how to consistently solve this issue...And this is just 1 of MILLIONS of problems they have ZERO idea how to solve.
All the driving by that sign with the computer on collecting data every day will NOT get you a result out of the ML/AI that works, because more data doesn’t solve that problem.
Yes ML more data helps with training generally.. but it DOESN’T in every case... and it WON’T in every case, and that’s my point.
A truly safe autonomous vehicle is DECADES away, and the reality is, is that most likely to be SAFE the default behavior is going to be drive slow, and when confused STOP. Meaning that a truly SAFE autonomous vehicle is likely going to be VERY niche.. IE people with physical disabilities etc that prevent them from driving they will be a God Send for their quality of life... to be fully independent.. .even if it means it may take them 45 minutes to commute what would take you 15.
I am all for autonomous vehicles, I am, I hope they have them ready by the time I am old so I do not have to find myself relying on others to get around.
With that said however, the problems that need to be solved to get there are huge in number, and more field data isn’t going to solve them. Its going to take major leaps in a lot of areas to get there.. because these aren’t oh, more data will get us an answer.. problems they are “WE HAVE ZERO IDEA HOW TO GET TO AN ANSWER” problems... and the more miles they put on the roads with these vehicles the list of problems they don’t know how to solve just keeps getting bigger, not smaller.
Eventually it will get there, but anyone telling you its close is LYING TO YOU, or WHOLESALE IGNORANT. Its DECADES away
I have never thought self-driving cars were feasible in today's age. The only way I see them working is if the traffic grid is rebuilt as something like a giant slot-car race track. The track would essentially do the driving. The only other way is if there is some sort of advanced android driver like Commander Data from Star Trek at the wheel.
A agree completely.
The human brain is amazing, they designed computers and logic systems, yet even we make mistakes.
Thats the fundamental problem with their business model. In and of itself the simple concept of ordinary people providing transport is a sound one, as demonstrated by the popularity of Uber. However, they have spent too much on R&D on things like autonomous vehicles etc. Thats where theyve p*ssed away their money.
Ahhh the tyranny of numbers..
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.