Short sellers - Cut your losses before the year end

[busarider29]

In the last few days, there were 3 crashes with 3 deaths which may related to the self-driving system.

Crashes and resulting fatalities are never acceptable. However, there are going to be mishaps and bumps along the road as we perfect these things (self-driving cars). It's inevitable. Because there is a crash or even a fatality resulting in a failure of a self-driving car, or any technology for that matter, doesn't mean we stop testing and perfecting. Will we see another report of a crash and/or fatality from a failure of an automated system? Yep, no doubt about it! But if we wish to perfect the automated driving system and road network, we have to be willing to go through the growing pains of developing and perfecting it. The eventual paybacks in lives saved and reduced accidents will be well worth the sacrifices made to develop the systems.

 

[harryting]

I just don't get the need for self-driving, RIGHT NOW. I'm sure others in the forum do it too, that during your daily driving, look around and think, huh, how would a automated car handle that situation? To me, it would be wise to wait until AI becomes better before pushing on with it.

Trucks on highway between inter-city-depots is a different matter though. There is high demand, today, for that. Even though it will be a huge social disrupter, being that truck driving is the largest employer for men without a college degree in US.

 

[Elcan]

I just don't get the need for self-driving, RIGHT NOW. I'm sure others in the forum do it too, that during your daily driving, look around and think, huh, how would a automated car handle that situation?

Elder people, drunk people, sleepy/tired people, people with severe disabilities or sick, etc.. All of them (and the rest of the people) will benefit from self-driving. Even people in good conditions make fatal errors.

To me, it would be wise to wait until AI becomes better before pushing on with it.

I understand that you don't want people dying while the technology is not very proved but, like somebody else said here, every (new) technology has its risks. Should we stop driving because there are car accidents? And cars have been around for more than a century.

 

[busarider29]

I just don't get the need for self-driving, RIGHT NOW.

If not now, when??? The develpments in AI are there now. We can do this, and we are. I can’t even begin to tell you how many times I have witnessed or read about an automobile accident and thought to myself, “An automated car and/or road network likely would have prevented that”. There should be no question there is a need for it NOW!! Just the other day, a car drove up beside me at a stop light and literally rammed the car right in front of it that was stopped. Likely the person driving wasn’t paying attention and/or looking at the phone or whatever. An automated car would have prevented that, no doubt. How about that accident in Indiana where the young lady wiped out three little siblings crossing the road to get on the school bus?!? Killed all three – two little twin brothers and their older sister. The bus was stopped with lights flashing and its Stop sign out. The young lady driving the truck that wiped out the kids simply wasn’t paying attention. Tell me again we don’t need automated cars and road networks RIGHT NOW. You and I read about and witness stuff like this every day, but some of us just accept it. Its time to develop the automated car and road network…….now!
Obviously, the driving force for the automated driving system is increased safety and less accidents, and that’s great. But what about all the other Nth-order benefits from automated cars that we’ll get, some of which we can’t even foresee right now. Lets just talk tires for instance. With an automated car and road network, we’ll go through a lot less tires because the automated car will drive much more efficiently, not slamming on the brakes, not slamming on the accelerator, not taking turns so fast, etc. So less tire use means more $$ in my pocket for other things, less oil use nation and world wide, less pollution, etc, etc. And the benefits of those benefits continue on and on and on. And we’re just talking about tires!! Furthermore, one day you’re not going to be able to drive a car, whether you like it or not. That’s just a fact of life. I’ll bet you’ll be all for automated cars and road networks then.

 

[Peter Nachtwey]

OT but I had too

I just don't get the need for self-driving, RIGHT NOW.

I see self driving as the first step to traffic management. People are notoriously bad at cooperative driving and it only takes a few to really screw up traffic. Some people are too cautious. Some are too aggressive. I hate driving in the city because there are so many idiots slowing down traffic. Even those that go fast slow down traffic because they expect others to make way for them thus slowing them down.


Imagine having a super computer that coordinates traffic on the main highways in cities. The easiest way to do this is to assign cars to a line of cars, each following the other at the optimal distance. Not too close or far behind the car in front. The car's computer could give the optimizing computer specs on how well the car can stop.


The goal of the optimizing computer is to optimize car flux. That is maximizing the flow over key sections of road. Take any section of road. If more cars enter a section of road than leave then it becomes congested. With today's technology it is easy to keep track of this. The optimizing computer would need to monitor the cars entering and leaving a section of road trying to maintain optimal flux/flow. I don't mean speed exactly but you want cars to exit at optimal rates. This can be measured at cars per minute. So for every section of road the optimizing computer would keep track of how many cars are entering and leaving a section of road per minute then integrate that to get the number of cars then divide by the length of road to get the density.


Obviously the optimizing computer would need to know where the car is going so it car predict a bit. The optimizing computer would also need to know how to control the cars to optimize traffic.


I can't see where this is so hard when Alpha Star can play very complicated games such as StarCraft II. I have a computer with a GTX 1080 graphics card. I didn't get it for graphics but for the 2560 CUDA cores that allow massively parallel computing for playing chess.


I hate human drivers. They are so stupid. How many times do you see cars race up to a red light only to stop. I try to coast up to red lights and estimate when they turn green so I have the maximum kinetic energy while going through the green light while the other dummies must accelerate from a stop. If I get stuck behind these dummies, I must stop too and accelerate from a stop. I could do a better job of maximizing my speed through green lights if I knew when the light was going to turn green.


Another problem I see is that cars accelerate much more slowly than they decelerate. That means if there is a delay, cars will accumulate at the back of the jam faster than the cars in front will leave the congested area. The car in front thinks it is saving money by not accelerating too fast but the car in front is essentially wasting the fuel of everyone behind them. I think we spend more time behind other cars than in front.



I think the means of optimizing traffic flow is just waiting for the ability to control the cars.

 

[drbitboy]

Or maybe, figure out a way for the EV car to charge while it is driving? So it's always self-charging on the road.

That's a heck of a long extension cord, maybe every buys a 1-mile long extension cord, and we invent AI-driven to trade connections as we pass by each other



or this

 

[drbitboy]

that should be "... we invent an AI-driven robot ..."

 

[drbitboy]

AI Driving is analogous to Nuclear Power

>Originally Posted by harryting View Post
>I just don't get the need for self-driving, RIGHT NOW.

If not now, when??? [...]

This is analogous to the coal/fossil vs. nuclear power choice: coal will kill far more people per kWh than nuclear (power) ever will, whether given historical or some rational number of nuclear accidents; coal also releases far more radiation into the environment, along with many other pollutants, than nuclear ever will. The volume of coal ash is probably as many orders of magnitude more than nuclear waste per kWh than nuclear waste is more toxic than coal ash, and having a smaller volume of waste makes it easier to deal with. A 650MW nuclear core is about 12feet high and 12 feet in diameter, and the toxic portions can be concentrated, so even allowing for multiple cores over a single plant's life cycle there will never be a "mountain" of nuclear waste.

But those facts are (unfortunately) irrelevant, because what it boils down to is this:

Some people won't accept the risk of death from a nuclear accident; they prefer a higher but more understandable death rate from coal.

Some people won't accept the of risk death from an AI-driven vehicle; they prefer a higher but more understandable death rate from poor drivers like themselves.

It's basically the plot from the movie [I, Robot].

 

[jimtech67]

AI sounds good..
Until AI starts making the decisions ...

Ask Sarah Connor

 

[alive15]

I believe whenever the media starts pushing EV vehicles and the benefits, the pros, etc, then it will become widespread. I don't think humans alone will ever just change, something major in the media must happen, like CNN, Fox, maybe Youtube.

 

[drbitboy]

I believe whenever the media starts pushing EV vehicles and the benefits, the pros, etc, then it will become widespread. I don't think humans alone will ever just change, something major in the media must happen, like CNN, Fox, maybe Youtube.

The media business model is pimping our eyes to the next commercial. They have no interest in EVs, or anything else for that matter, whatsoever, other than how they can serve that business model i.e. deliver eyes to the next commercial, or more specifically to the as-measured proxy for the next commercial i.e. Nielsen Ratings.

The most likely way to make that happen is news stories about people hurt or killed by AI-driven cars.

 

[drbitboy]

pimping eyes

pimping our eyes to the next commercial

To be clear about this analogy:

• TV-watcher's eyes are the prostitute i.e. the product
• the next commercial, or the sponsor of same, is the john i.e. the one with the money
• The network is the pimp, the middle-man who ends up with the the money from the john and the "owners" of the product

The networks could care less about what is broadcast between the commercials, other than that it serve the business model.

Explains a lot, eh?

 

[harryting]

Elder people, drunk people, sleepy/tired people, people with severe disabilities or sick, etc.. All of them (and the rest of the people) will benefit from self-driving. Even people in good conditions make fatal errors.

I understand that you don't want people dying while the technology is not very proved but, like somebody else said here, every (new) technology has its risks. Should we stop driving because there are car accidents? And cars have been around for more than a century.

That's reading a bit too much into what I said. People die everyday, there's a higher risk of my dying every time I leave the house, I get it.

I feel there's a lot more to the "brainless" driving than we realize. There are scenarios that would involve making "ethical" decision. Like running over a blown up toy vs a small child vs a deer.

I hope to be wrong but my prediction is that legal matter would prevent the adoption of it more than technology. Which means this may happen sooner in a authoritarian nation first.

Also, does it really hurt to wait? To draw an analogy to interstellar travel, on a podcast I heard. That we might we well not try to put a craft out now and just wait for better technology because the craft with latter technology will actually fly right by the older spacecraft and get to its destination first.

I'm saying we will probably end up with fully autonomous surface travel around the very same time, even if we wait for better technology to come along.

Oh, btw. Nuclear IS the safest way to generate power (that is, even taking account the few famous accidents).

 

[Peter Nachtwey]

Oh, btw. Nuclear IS the safest way to generate power (that is, even taking account the few famous accidents).

I agree.

I an an ex-nuke. I was reactor control officer on a nuclear submarine.
However, the power plants must be the right kind. I am not a fan of the kind the Russians had at Chernobyl. I also thing the Dai-ichi plants were a poor design because radioactive water would run through the turbines.
Naval reactors have a primary and secondary coolant system. The primary and secondary coolants are separated by a steam generator. The steam generator is the weak point of these pressurized water reactors. If there is a leak between the primary and the secondary water then the secondary gets contaminated. This is what happened to our local Trojan nuclear reactor. Leaks can occur due to poor water chemistry. Water chemistry is crucial for minimizing corrosion.

There can be no short cuts and there must be better training than those people at 3 mile island got.


BTW, I was interviewed by Adm Rickover twice.

 

[dmroeder]

While nuclear power has proven very safe statistically, the consequences of a failure are just so bad. Even without a failure, the material and waste is so toxic and toxic for such long periods of time. I'm not anti-nuke, but I can understand the hesitation towards using it. I know there are efforts in using different isotopes as well as using the waste from reactors as fuel for newly designed reactors, that would be amazing. Whatever it takes to reduce the toxicity of the waste and the half-life.

Now if we could only have an "arms race" on fusion. A lot of people are negative on fusion, but it's really just an engineering problem. I'm sure fusion could be worked out with a Manhattan Project style effort.

This topic has gone off the rails a bit hasn't it?