Self-driving Chevy Bolts are coming to the streets of Manhattan in 2018

Its mean streets will be the self-driving car's toughest challenge.

Read the whole story

 

[Foiler]

I realize my opinion may not be popular, especially on a tech news site, but I just don't like where all this is going. It can be argued either way, but the side I am on is that this will ultimately dumb down our society. I refer to the movie "Wall-E" countless times these days as I think into what the future might look like. There will one day be a generation that has no idea how to drive a vehicle 'manually'. Think "Demolition Man"...

Businesses make stuff to sell to people that want to buy their stuff. Simple. So this kind of thing will be a 'success' based on that statement alone. The sale is easy once a person sees that they can do other stuff while in the car; especially the ones that have long commutes to work daily (like myself). I did the math a while ago, and I literally spend 1 month each year just driving to and from work. It is appealing to me to get one of these cars; I won't deny it.

Let's just say there will NEVER be a wide-spread hack of Tesla's network, and let's just say there will NEVER be an accident caused by an AV. Because none of that will EVER happen. /s

Study this and study that, I get it. They are safer. We can't deny it. They react much faster than humans do; more specifically the distracted human. I believe safety will also be a huge factor into the success story of the AV.

Although the AV concept has already left the station (punny), I believe the focus should be shifted to educating and enforcing safe driving. Using a smart phone while driving should be equivalent to drunk driving. Think about it -- they both weave in and out of their lane, and they both cause thousands of fatalities each year.

Do we really need this?

Yes, we need this. 38,000 people died on US roads in 2015. 100 people per day. Not to mention something like 4 million serious injuries due to traffic accidents each year as well. It's unbelievable that we tolerate this. I don't know why we are so de-sensitized to it.

Could you imagine getting a text every time someone was injured in a car accident? It'd be something like every 10 seconds... "broken arm", "broken nose", "fatality", over and over and over and over again.

I appreciate your optimism about humans and driving, but everyone knows that texting and driving is dangerous just like everyone knows that driving drunk is dangerous. But still people do it. And people fall asleep while driving. And people look back into the back seat while driving. Or put on their makeup. Or pick up something off the floor... or take their eyes off the road for just a moment...

There will no doubt be significant displacement because of the self-driving car revolution. But 30,000 dead people year after year after year? 4 million serious injuries year after year after year? I know which side I'm on.

The desensitized reason as you say is because the numbers have grown steadily as the population increases. In 10 years you could say it even more-so. The real question is: Have the percentages gone down because of all of these safety regulations? If there were a trillion people living on this planet, you would think the deaths were even more staggering even if it were only a 0.01% death rate. We also consider it their own fault for accidents. It's either negligence or not paying attention, or that they follow too close and don't give themselves enough reaction time. They can only blame themselves. For these self-driving cars - it is the computer/companies fault for the error/wreck. The real issue is when one of these self-wrecking cars malfunctions and someone gets a one-way ride off a bridge or any other of the trillions of daily variables. Also, these cars are only reactive and not proactive. They follow a car down the highway at 50mph. It can't see that the tree is about to fall into the road. A human would see this and stop. The computer wouldn't think there was a problem until the tree has already fallen on it. Same thing with furniture or rocks coming off of a trailer. Same thing with a rock rolling off a hill or, like I said, any of the other trillion variables. Not to mention the sensors and cameras failing, computer errors, etc etc. People don't keep their vehicles in perfect condition or maintenance either. If they don't do oil changes do you think they are going to spend the thousands of $$ to keep it working? Not to mention software updates never happen like TVs and blu-ray players. There is a long long road to go before these should be put in production. No pun intended.

 

[hutsell]

Fta: Manhattan will be an even tougher challenge for the robocars. Manhattan's roads are a hellish agglomeration of potholes, double- and even triple-parking, and pedestrian and vehicle traffic unlike anywhere else in the country.

Not to forget or ignore the reflectivity of falling snow and the snow/slushed covered streets during the winter season -- a condition existing in 30 States with about 60% of the US population.

A quick informal search gave a few titles implying it's been solved until I actually waded through some of the articles saying otherwise; there are solutions but they still need improvement. Unless I've missed something, it still appears to be a fair-weather technology ideally suited for most places during the summer or throughout the year anywhere in the southwest or Hawaii.

Then again, if NYC has some type of plowing process on steroids, it's been considered a non-issue(?).

 

[CommentatorForNow]

I realize my opinion may not be popular, especially on a tech news site, but I just don't like where all this is going. It can be argued either way, but the side I am on is that this will ultimately dumb down our society. I refer to the movie "Wall-E" countless times these days as I think into what the future might look like. There will one day be a generation that has no idea how to drive a vehicle 'manually'. Think "Demolition Man"...

Businesses make stuff to sell to people that want to buy their stuff. Simple. So this kind of thing will be a 'success' based on that statement alone. The sale is easy once a person sees that they can do other stuff while in the car; especially the ones that have long commutes to work daily (like myself). I did the math a while ago, and I literally spend 1 month each year just driving to and from work. It is appealing to me to get one of these cars; I won't deny it.

Let's just say there will NEVER be a wide-spread hack of Tesla's network, and let's just say there will NEVER be an accident caused by an AV. Because none of that will EVER happen. /s

Study this and study that, I get it. They are safer. We can't deny it. They react much faster than humans do; more specifically the distracted human. I believe safety will also be a huge factor into the success story of the AV.

Although the AV concept has already left the station (punny), I believe the focus should be shifted to educating and enforcing safe driving. Using a smart phone while driving should be equivalent to drunk driving. Think about it -- they both weave in and out of their lane, and they both cause thousands of fatalities each year.

Do we really need this?

Yes, we need this. 38,000 people died on US roads in 2015. 100 people per day. Not to mention something like 4 million serious injuries due to traffic accidents each year as well. It's unbelievable that we tolerate this. I don't know why we are so de-sensitized to it.

Could you imagine getting a text every time someone was injured in a car accident? It'd be something like every 10 seconds... "broken arm", "broken nose", "fatality", over and over and over and over again.

I appreciate your optimism about humans and driving, but everyone knows that texting and driving is dangerous just like everyone knows that driving drunk is dangerous. But still people do it. And people fall asleep while driving. And people look back into the back seat while driving. Or put on their makeup. Or pick up something off the floor... or take their eyes off the road for just a moment...

There will no doubt be significant displacement because of the self-driving car revolution. But 30,000 dead people year after year after year? 4 million serious injuries year after year after year? I know which side I'm on.

The desensitized reason as you say is because the numbers have grown steadily as the population increases. In 10 years you could say it even more-so. The real question is: Have the percentages gone down because of all of these safety regulations? If there were a trillion people living on this planet, you would think the deaths were even more staggering even if it were only a 0.01% death rate. We also consider it their own fault for accidents. It's either negligence or not paying attention, or that they follow too close and don't give themselves enough reaction time. They can only blame themselves. For these self-driving cars - it is the computer/companies fault for the error/wreck. The real issue is when one of these self-wrecking cars malfunctions and someone gets a one-way ride off a bridge or any other of the trillions of daily variables. Also, these cars are only reactive and not proactive. They follow a car down the highway at 50mph. It can't see that the tree is about to fall into the road. A human would see this and stop. The computer wouldn't think there was a problem until the tree has already fallen on it. Same thing with furniture or rocks coming off of a trailer. Same thing with a rock rolling off a hill or, like I said, any of the other trillion variables. Not to mention the sensors and cameras failing, computer errors, etc etc. People don't keep their vehicles in perfect condition or maintenance either. If they don't do oil changes do you think they are going to spend the thousands of $$ to keep it working? Not to mention software updates never happen like TVs and blu-ray players. There is a long long road to go before these should be put in production. No pun intended.

The only thing that really matters to me is overall safety, and I have absolute confidence that over the next few years, this tech will become safer than human drivers. It's not a matter of if but when.

I do think you have articulated some hurdles to getting the technology adopted... in particular the misguided sense that the average human driver is paying attention. In your scenario, you act as if every human driver is going to be paying attention when the tree is falling (and I actually think most AV technologies will be in a place to respond to this in the next couple of years). But humans often are not paying attention. I think the hardest part about driving is not interrupting your attention. And that's an promise AVs can keep, and humans cannot.

 

[Chuckstar]

As much as I love technological progression, (who honestly wants to go back to mid-90s tech?), I'm afraid that parts of society's progression will drag behind technological progression and we may experience some severe growing pains. Not enough jobs to go around coupled with more and more jobs being automated out of existence with a heavy dose of "You don't work, you don't eat, fuck you" mindset, which takes about two seconds to find examples of, would be pure pain and misery for countless number of people.That worries me.

I too, am no Luddite, but I can easily see that this kind of automation will cause societal havoc. Millions of jobs will be lost world wide, and this is just the beginning. I don't welcome this at all.

I'm not FOND of the idea, but your alarmist views fail to take into account a little something called "history".

Millions of jobs will be CREATED when you lose millions of jobs. The fact is no matter what industry you want to look at, once you get a paradigm shift in that industry (transportation going from horse to automobile, for example), jobs are ALWAYS lost. Blacksmiths and livery stables became non-viable due to vast oversupply of those services in the economy.

But a major increase in job availability came as a result of new road construction, blacksmiths became car mechanics, livery stables became gas stations and everyone had jobs again.

That's just ONE example of how an economy handles changes in technology. It's happened time and again when people simply find new vocations/occupations to handle changes in how things are done.

So the whole doom and gloom, "the sky is falling" and "millions will be unemployed" nonsense is, well, pure bullshit from an overall point of view. The jobs will be there in some form or another.

I hope you are correct, but I am dubious that this reference to historical trends can be projected forward in perpetuity.

Your examples are entirely valid, but they aren't exactly always relevant.

Technology shifts jobs in one of two ways: It can change the fundamental demand for goods and services, and it can shift workers from one labor "category" to another.

Let's start with your example of the transition from horse-driven transportation to automobiles. What changed here was a change in demand from one type of good to another (horses and their supporting infrastructure to automobiles and their supporting infrastructure). The reason there was a net gain in jobs was because of two factors:

1) The labor category didn't change. We had human beings creating things and human beings offering services for both environments. The jobs just transitioned from one product to another.

2) The demand increased because of the overall increase in material wealth and the complexity of providing the services. Cars are much more labor intensive to have as part of our economy than horses. They need to be manufactured, serviced, fueled and require an extensive road network. All of this leads to a greatly increased need for a labor pool to support the ecosystem. Ergo: even more jobs.

But as automation starts to enter the picture, many of the jobs that used to be taken up by humans start to vanish. Things like manufacturing, road work, etc. (For example, look at coal mining. The loss of jobs there is as much from automation as it is from a reduction in demand). These jobs are permanently lost to automation. But until now, these lost jobs are more than made up for by the increased number of jobs in information and services.

This is what I mean by a transition from one labor 'category' to another. Manufacturing and other easy to automate jobs are lost, but the humans shift to the next labor categories: Information workers and service workers. And as the economy grows, the demand for these types of goods and services also grows.

But the problem here is that there is not an endless supply of labor categories, and there is not an endless opportunity for growth. As automation becomes more sophisticated, people will be displaced from service work (for example: drivers replaced by automated vehicles). These people cannot (generally) switch back to manual labor jobs because those have already been displaced. They will have to switch to information work. But as automation starts to encroach there (think IBM's Watson) there will be fewer jobs available there as well.

Meanwhile, the implication that an ever growing economy can drive enough increased demand to offset this loss of jobs is not reliable anymore either. It is much easier to absorb this increased growth with more automated jobs than with more, highly trained humans. So as demand for new types of goods and services grows, there will no longer automatically be a corresponding increase in demand for labor to provide these goods and services. That is already evident today. Even as we see whole new industries develop, we aren't seeing massive increases in any jobs in areas that are already automated. We only see increases in those labor categories that are still filled by humans. But once a category is automated, even the largest increases in demand in that category will not result in new jobs.

And we are running out of labor categories.

But there is an even simpler way of looking at it. The closer technology becomes to doing what a human can do, it will always be easier and cheaper to double the number of these 'robots' than it will be to add more humans to the labor pool.

People have been making this argument since they first started using draft animals. "Now that oxen can pull our carts, we'll need so many fewer workers."

There's no such thing as "running out of labor categories". That's just nonsense.

 

[Chuckstar]

I'd suspect Manhattan would probably be easier. Especially in the gridded sections. The roads are pretty straightforward, and speed limits are relatively low at 25MPH.

The toughest challenge would probably be pedestrians, but pedestrians usually stay out of the car's way.

Haha, never been to New York? I wanna see how this deals with specific intersections like anything along Times Square where you need to inch up until the point where the pedestrians decide to go behind you instead of in front of you. And only then are you able to actually go through the intersection.

Or the grid lock version where if you don't force your way into the intersection, the opposing traffic will eventually block you out again once they get the green light.

Or that often during busy times there's a policeman directing traffic in some intersections. Do autonomous cars know how to follow police hand signals, yet?

However, I understand there are drivers in the cars, so such things won't actually end up being a problem. Probably these are the kinds of reasons they want to test somewhere like Manhattan. Not because they've necessarily solved the problems, but because the only way to really solve them is by getting real-world data to work with.

 

[Fatesrider]

As much as I love technological progression, (who honestly wants to go back to mid-90s tech?), I'm afraid that parts of society's progression will drag behind technological progression and we may experience some severe growing pains. Not enough jobs to go around coupled with more and more jobs being automated out of existence with a heavy dose of "You don't work, you don't eat, fuck you" mindset, which takes about two seconds to find examples of, would be pure pain and misery for countless number of people.That worries me.

There is no reason to think automation reduces employment on a national scale. It changes what jobs there are and what they pay. It increases productivity, and improves quality in the industries where it is applied. But there is no evidence that on a national scale it leads to unemployment. And there is no reason to think that banning it or freezing it will lead, nationally, to better outcomes.

It may be that if you are an old fashioned print operator, the move to electronic typesetting lowers your premium and your wages. But there is no reason to think that the newspaper industry moving to electronics lowers employment for the whole country. In fact, it probably increases it. But it does lower the number of typesetters.

However, as we can see with the coal industry in Appalachia, destroying the jobs is easy. Getting the people retrained and employed is a different issue altogether, especially the older the worker is.

There's nothing new about that, though.

And it very much depends on the individual. A lot of smiths didn't care about horses so much as the metal. Body shop mechanic was just fine, even if it was an entirely new skill-set.

The other point is that generational family trades are very much a thing of the past. That whole tradition is fraught with job stability peril these days (and makes exactly as much sense as "brand loyalty"). The employment market is still basically a slave mentality these days, so if you're not proactive in getting out there and diversifying your skill set (whether before your job goes extinct due to tech or lack of demand, or after), it's pretty much on you to suffer the consequences of your lack of motivation.

The job industry doesn't care what warm body is placed in that particular cog's position. Be it you or someone else. So, if one needs a job, do what one needs to do to get it and be the cog. OTOH, one could try to raise money out of selling coal chunks (say covered in a clear resin) for fun stocking stuffers for the holidays and send them to the administration who said they'd bring your jobs back.

Considering how many Americans would love to send this administration a lump of coal for Christmas, that'd probably pay well enough to go to an ivy league school.

As for autonomous vehicles on the streets of Manhattan, I simply shrug. It's a test. That's all. It will be in restricted zones under specific conditions and hardly a "life fire" exercise, no matter what they say. The program will close, they'll look at the data and fix what went wrong and try again. They have a very long way to go before this tech is ready for prime time like cars are today.

It's all part of the process.

 

[bvz_1]

As much as I love technological progression, (who honestly wants to go back to mid-90s tech?), I'm afraid that parts of society's progression will drag behind technological progression and we may experience some severe growing pains. Not enough jobs to go around coupled with more and more jobs being automated out of existence with a heavy dose of "You don't work, you don't eat, fuck you" mindset, which takes about two seconds to find examples of, would be pure pain and misery for countless number of people.That worries me.

I too, am no Luddite, but I can easily see that this kind of automation will cause societal havoc. Millions of jobs will be lost world wide, and this is just the beginning. I don't welcome this at all.

I'm not FOND of the idea, but your alarmist views fail to take into account a little something called "history".

Millions of jobs will be CREATED when you lose millions of jobs. The fact is no matter what industry you want to look at, once you get a paradigm shift in that industry (transportation going from horse to automobile, for example), jobs are ALWAYS lost. Blacksmiths and livery stables became non-viable due to vast oversupply of those services in the economy.

But a major increase in job availability came as a result of new road construction, blacksmiths became car mechanics, livery stables became gas stations and everyone had jobs again.

That's just ONE example of how an economy handles changes in technology. It's happened time and again when people simply find new vocations/occupations to handle changes in how things are done.

So the whole doom and gloom, "the sky is falling" and "millions will be unemployed" nonsense is, well, pure bullshit from an overall point of view. The jobs will be there in some form or another.

I hope you are correct, but I am dubious that this reference to historical trends can be projected forward in perpetuity.

Your examples are entirely valid, but they aren't exactly always relevant.

Technology shifts jobs in one of two ways: It can change the fundamental demand for goods and services, and it can shift workers from one labor "category" to another.

Let's start with your example of the transition from horse-driven transportation to automobiles. What changed here was a change in demand from one type of good to another (horses and their supporting infrastructure to automobiles and their supporting infrastructure). The reason there was a net gain in jobs was because of two factors:

1) The labor category didn't change. We had human beings creating things and human beings offering services for both environments. The jobs just transitioned from one product to another.

2) The demand increased because of the overall increase in material wealth and the complexity of providing the services. Cars are much more labor intensive to have as part of our economy than horses. They need to be manufactured, serviced, fueled and require an extensive road network. All of this leads to a greatly increased need for a labor pool to support the ecosystem. Ergo: even more jobs.

But as automation starts to enter the picture, many of the jobs that used to be taken up by humans start to vanish. Things like manufacturing, road work, etc. (For example, look at coal mining. The loss of jobs there is as much from automation as it is from a reduction in demand). These jobs are permanently lost to automation. But until now, these lost jobs are more than made up for by the increased number of jobs in information and services.

This is what I mean by a transition from one labor 'category' to another. Manufacturing and other easy to automate jobs are lost, but the humans shift to the next labor categories: Information workers and service workers. And as the economy grows, the demand for these types of goods and services also grows.

But the problem here is that there is not an endless supply of labor categories, and there is not an endless opportunity for growth. As automation becomes more sophisticated, people will be displaced from service work (for example: drivers replaced by automated vehicles). These people cannot (generally) switch back to manual labor jobs because those have already been displaced. They will have to switch to information work. But as automation starts to encroach there (think IBM's Watson) there will be fewer jobs available there as well.

Meanwhile, the implication that an ever growing economy can drive enough increased demand to offset this loss of jobs is not reliable anymore either. It is much easier to absorb this increased growth with more automated jobs than with more, highly trained humans. So as demand for new types of goods and services grows, there will no longer automatically be a corresponding increase in demand for labor to provide these goods and services. That is already evident today. Even as we see whole new industries develop, we aren't seeing massive increases in any jobs in areas that are already automated. We only see increases in those labor categories that are still filled by humans. But once a category is automated, even the largest increases in demand in that category will not result in new jobs.

And we are running out of labor categories.

But there is an even simpler way of looking at it. The closer technology becomes to doing what a human can do, it will always be easier and cheaper to double the number of these 'robots' than it will be to add more humans to the labor pool.

People have been making this argument since they first started using draft animals. "Now that oxen can pull our carts, we'll need so many fewer workers."

There's no such thing as "running out of labor categories". That's just nonsense.

I'm not sure it is.

What labor categories can you think of? Remember, a labor category isn't "cobbler" vs. "machinist". It is physical labor vs. service vs. information worker, etc.

The reality is that we automate from the least "decision" oriented work to the most. Someone who just cranks a crank at a constant pace can easily be replaced by a motor because there is no decision making capability involved.

Early automation that involved "decisions" could be handled by trip switches etc. Then there was some computerization and more jobs switched over from being supplied by humans to being automated. We compensated, like I pointed out in my previous post, by having more jobs that require complex decision making, and a larger economy that relies on these types of labor more than it ever has.

But the level of automation of complex decision making is increasing at an ever increasing rate. When a computer can make a decision as capably (or better, or even just cheaper) than a human, then that whole class of labor is permanently lost to automation.

All of the examples of automation in the past have been with extremely limited decision making capabilities with humans just climbing the ladder as the rungs disappear behind them. That is what I meant by shifting job 'categories'. As the capability of automated decision making becomes more complex there is less and less room at the top of the ladder. There simply aren't enough new areas of complex decision making left to us to all crowd into. Jobs will be irrevocably lost.

 

[icwhatudidthere]

I'd suspect Manhattan would probably be easier. Especially in the gridded sections. The roads are pretty straightforward, and speed limits are relatively low at 25MPH.

The toughest challenge would probably be pedestrians, but pedestrians usually stay out of the car's way.

Haha, never been to New York? I wanna see how this deals with specific intersections like anything along Times Square where you need to inch up until the point where the pedestrians decide to go behind you instead of in front of you. And only then are you able to actually go through the intersection.

Or the grid lock version where if you don't force your way into the intersection, the opposing traffic will eventually block you out again once they get the green light.

Or that often during busy times there's a policeman directing traffic in some intersections. Do autonomous cars know how to follow police hand signals, yet?

However, I understand there are drivers in the cars, so such things won't actually end up being a problem. Probably these are the kinds of reasons they want to test somewhere like Manhattan. Not because they've necessarily solved the problems, but because the only way to really solve them is by getting real-world data to work with.

True true, and good point about traffic police. The cars will need to be able to handle a popup traffic cop.

Also, what to do when a traffic light is just dead. A few weeks ago on 14th and 1st, the lights were just off. No traffic cop had taken over yet and power seemed to be completely out so no flashing red or yellow. So it was just drivers very carefully approaching the intersection and negotiating with other drivers.

I do suspect however that the day after these rollout, we'll have the answer to what happens in a game of chicken between a pedestrian and an autonomous vehicle.

 

[Chuckstar]

As much as I love technological progression, (who honestly wants to go back to mid-90s tech?), I'm afraid that parts of society's progression will drag behind technological progression and we may experience some severe growing pains. Not enough jobs to go around coupled with more and more jobs being automated out of existence with a heavy dose of "You don't work, you don't eat, fuck you" mindset, which takes about two seconds to find examples of, would be pure pain and misery for countless number of people.That worries me.

I too, am no Luddite, but I can easily see that this kind of automation will cause societal havoc. Millions of jobs will be lost world wide, and this is just the beginning. I don't welcome this at all.

I'm not FOND of the idea, but your alarmist views fail to take into account a little something called "history".

Millions of jobs will be CREATED when you lose millions of jobs. The fact is no matter what industry you want to look at, once you get a paradigm shift in that industry (transportation going from horse to automobile, for example), jobs are ALWAYS lost. Blacksmiths and livery stables became non-viable due to vast oversupply of those services in the economy.

But a major increase in job availability came as a result of new road construction, blacksmiths became car mechanics, livery stables became gas stations and everyone had jobs again.

That's just ONE example of how an economy handles changes in technology. It's happened time and again when people simply find new vocations/occupations to handle changes in how things are done.

So the whole doom and gloom, "the sky is falling" and "millions will be unemployed" nonsense is, well, pure bullshit from an overall point of view. The jobs will be there in some form or another.

I hope you are correct, but I am dubious that this reference to historical trends can be projected forward in perpetuity.

Your examples are entirely valid, but they aren't exactly always relevant.

Technology shifts jobs in one of two ways: It can change the fundamental demand for goods and services, and it can shift workers from one labor "category" to another.

Let's start with your example of the transition from horse-driven transportation to automobiles. What changed here was a change in demand from one type of good to another (horses and their supporting infrastructure to automobiles and their supporting infrastructure). The reason there was a net gain in jobs was because of two factors:

1) The labor category didn't change. We had human beings creating things and human beings offering services for both environments. The jobs just transitioned from one product to another.

2) The demand increased because of the overall increase in material wealth and the complexity of providing the services. Cars are much more labor intensive to have as part of our economy than horses. They need to be manufactured, serviced, fueled and require an extensive road network. All of this leads to a greatly increased need for a labor pool to support the ecosystem. Ergo: even more jobs.

But as automation starts to enter the picture, many of the jobs that used to be taken up by humans start to vanish. Things like manufacturing, road work, etc. (For example, look at coal mining. The loss of jobs there is as much from automation as it is from a reduction in demand). These jobs are permanently lost to automation. But until now, these lost jobs are more than made up for by the increased number of jobs in information and services.

This is what I mean by a transition from one labor 'category' to another. Manufacturing and other easy to automate jobs are lost, but the humans shift to the next labor categories: Information workers and service workers. And as the economy grows, the demand for these types of goods and services also grows.

But the problem here is that there is not an endless supply of labor categories, and there is not an endless opportunity for growth. As automation becomes more sophisticated, people will be displaced from service work (for example: drivers replaced by automated vehicles). These people cannot (generally) switch back to manual labor jobs because those have already been displaced. They will have to switch to information work. But as automation starts to encroach there (think IBM's Watson) there will be fewer jobs available there as well.

Meanwhile, the implication that an ever growing economy can drive enough increased demand to offset this loss of jobs is not reliable anymore either. It is much easier to absorb this increased growth with more automated jobs than with more, highly trained humans. So as demand for new types of goods and services grows, there will no longer automatically be a corresponding increase in demand for labor to provide these goods and services. That is already evident today. Even as we see whole new industries develop, we aren't seeing massive increases in any jobs in areas that are already automated. We only see increases in those labor categories that are still filled by humans. But once a category is automated, even the largest increases in demand in that category will not result in new jobs.

And we are running out of labor categories.

But there is an even simpler way of looking at it. The closer technology becomes to doing what a human can do, it will always be easier and cheaper to double the number of these 'robots' than it will be to add more humans to the labor pool.

People have been making this argument since they first started using draft animals. "Now that oxen can pull our carts, we'll need so many fewer workers."

There's no such thing as "running out of labor categories". That's just nonsense.

I'm not sure it is.

What labor categories can you think of? Remember, a labor category isn't "cobbler" vs. "machinist". It is physical labor vs. service vs. information worker, etc.

The reality is that we automate from the least "decision" oriented work to the most. Someone who just cranks a crank at a constant pace can easily be replaced by a motor because there is no decision making capability involved.

Early automation that involved "decisions" could be handled by trip switches etc. Then there was some computerization and more jobs switched over from being supplied by humans to being automated. We compensated, like I pointed out in my previous post, by having more jobs that require complex decision making, and a larger economy that relies on these types of labor more than it ever has.

But the level of automation of complex decision making is increasing at an ever increasing rate. When a computer can make a decision as capably (or better, or even just cheaper) than a human, then that whole class of labor is permanently lost to automation.

All of the examples of automation in the past have been with extremely limited decision making capabilities with humans just climbing the ladder as the rungs disappear behind them. That is what I meant by shifting job 'categories'. As the capability of automated decision making becomes more complex there is less and less room at the top of the ladder. There simply aren't enough new areas of complex decision making left to us to all crowd into. Jobs will be irrevocably lost.

I can think of literally thousands of labor categories. Your artificial restriction of what a labor category is is nonsensical, and bears no relationship to how automation affects the workforce.

 

[ShagnWagn]

I realize my opinion may not be popular, especially on a tech news site, but I just don't like where all this is going. It can be argued either way, but the side I am on is that this will ultimately dumb down our society. I refer to the movie "Wall-E" countless times these days as I think into what the future might look like. There will one day be a generation that has no idea how to drive a vehicle 'manually'. Think "Demolition Man"...

Businesses make stuff to sell to people that want to buy their stuff. Simple. So this kind of thing will be a 'success' based on that statement alone. The sale is easy once a person sees that they can do other stuff while in the car; especially the ones that have long commutes to work daily (like myself). I did the math a while ago, and I literally spend 1 month each year just driving to and from work. It is appealing to me to get one of these cars; I won't deny it.

Let's just say there will NEVER be a wide-spread hack of Tesla's network, and let's just say there will NEVER be an accident caused by an AV. Because none of that will EVER happen. /s

Study this and study that, I get it. They are safer. We can't deny it. They react much faster than humans do; more specifically the distracted human. I believe safety will also be a huge factor into the success story of the AV.

Although the AV concept has already left the station (punny), I believe the focus should be shifted to educating and enforcing safe driving. Using a smart phone while driving should be equivalent to drunk driving. Think about it -- they both weave in and out of their lane, and they both cause thousands of fatalities each year.

Do we really need this?

Yes, we need this. 38,000 people died on US roads in 2015. 100 people per day. Not to mention something like 4 million serious injuries due to traffic accidents each year as well. It's unbelievable that we tolerate this. I don't know why we are so de-sensitized to it.

Could you imagine getting a text every time someone was injured in a car accident? It'd be something like every 10 seconds... "broken arm", "broken nose", "fatality", over and over and over and over again.

I appreciate your optimism about humans and driving, but everyone knows that texting and driving is dangerous just like everyone knows that driving drunk is dangerous. But still people do it. And people fall asleep while driving. And people look back into the back seat while driving. Or put on their makeup. Or pick up something off the floor... or take their eyes off the road for just a moment...

There will no doubt be significant displacement because of the self-driving car revolution. But 30,000 dead people year after year after year? 4 million serious injuries year after year after year? I know which side I'm on.

The desensitized reason as you say is because the numbers have grown steadily as the population increases. In 10 years you could say it even more-so. The real question is: Have the percentages gone down because of all of these safety regulations? If there were a trillion people living on this planet, you would think the deaths were even more staggering even if it were only a 0.01% death rate. We also consider it their own fault for accidents. It's either negligence or not paying attention, or that they follow too close and don't give themselves enough reaction time. They can only blame themselves. For these self-driving cars - it is the computer/companies fault for the error/wreck. The real issue is when one of these self-wrecking cars malfunctions and someone gets a one-way ride off a bridge or any other of the trillions of daily variables. Also, these cars are only reactive and not proactive. They follow a car down the highway at 50mph. It can't see that the tree is about to fall into the road. A human would see this and stop. The computer wouldn't think there was a problem until the tree has already fallen on it. Same thing with furniture or rocks coming off of a trailer. Same thing with a rock rolling off a hill or, like I said, any of the other trillion variables. Not to mention the sensors and cameras failing, computer errors, etc etc. People don't keep their vehicles in perfect condition or maintenance either. If they don't do oil changes do you think they are going to spend the thousands of $$ to keep it working? Not to mention software updates never happen like TVs and blu-ray players. There is a long long road to go before these should be put in production. No pun intended.

The only thing that really matters to me is overall safety, and I have absolute confidence that over the next few years, this tech will become safer than human drivers. It's not a matter of if but when.

I do think you have articulated some hurdles to getting the technology adopted... in particular the misguided sense that the average human driver is paying attention. In your scenario, you act as if every human driver is going to be paying attention when the tree is falling (and I actually think most AV technologies will be in a place to respond to this in the next couple of years). But humans often are not paying attention. I think the hardest part about driving is not interrupting your attention. And that's an promise AVs can keep, and humans cannot.

If you say you absolutely only care about overall safety, then you would be ok if you were the one it has an error with (software and/or hardware) and you get driven off a bridge? I would prefer my fate to be in my own hands, not some dude behind a desk trying to program these things.

 

[vdiv]

How many drivers does it take to operate a run-off-the-mill Bolt EV? One human.
How many drivers does it take to operate an autonomous Bolt EV? One computer and two humans.

Think we may have found the problem

 

[co-lee]

It sure looks to me like increasing automation is going to cause a lot of trouble for people wanting a job. And trouble in fields where a threat from automation seemed unthinkable not long ago. E.g, highly skilled professions like medicine are confronting the challenge when the computer does a better job diagnosing or reading the xray imagery or choosing the right medicine mixture. A lot of medical professionals are already basically moving into patient service and serving as gatekeepers for the insurance companies. This will only increase.

So, it's gonna be a big change.

Thinking that self-driving cars are the lynchpin of this change and that by holding onto your keys till they pry them from your cold dead hands is somehow going to preserve meaningful work is rather silly. It's more like the robot car is the canary in the coal mine, telling us that we're in for a rough ride. Giving the canary artificial air would't keep the miners from keeling over. And resisting self-driving cars as some kind of labor movement statement also isn't going to solve the problem.

We do need to work on how we support people with meaningful lives and work even as paid work becomes either unavailable or basically service. This is not going to be easy.

 

[bvz_1]

I can think of literally thousands of labor categories. Your artificial restriction of what a labor category is is nonsensical, and bears no relationship to how automation affects the workforce.

Ok, maybe you could list some of these? And maybe explain to me why my analysis is wrong (other than "it just is"). It may well be. I'm not an expert. But I'm trying to see where I am wrong beyond that I "just am".

I defined a labor "category" to be what it is so that I could describe how the hierarchy of complex decision making works. If you don't like that term, that is fine. We can call it something else. Let's call it the "complex decision making hierarchy".

As automation works its way up this hierarchy, there is no corresponding increase in complexity at the top end that we humans can crowd into. What broad "classes of complex decision making" jobs are out there still? Let's say that mechanical engineering gets automated. What kinds of jobs would these engineers go into that isn't similarly complex and, therefore, also likely to be automated by the same technology? That's what I meant by "categories". Once a level of complex decision making is automated, it is automated across the board at that level and we humans have to crowd into what remains. So far there has been enough space for most of us (numerically speaking), but that space is simply going to shrink more and more. There is an upper limit to the complex decision making that a human can do, and we are engineering machines that are encroaching in on the bottom end of that realm already.

 

[novelski]

I'd suspect Manhattan would probably be easier. Especially in the gridded sections. The roads are pretty straightforward, and speed limits are relatively low at 25MPH.

The toughest challenge would probably be pedestrians, but pedestrians usually stay out of the car's way.

Haha, never been to New York? I wanna see how this deals with specific intersections like anything along Times Square where you need to inch up until the point where the pedestrians decide to go behind you instead of in front of you. And only then are you able to actually go through the intersection.

Or the grid lock version where if you don't force your way into the intersection, the opposing traffic will eventually block you out again once they get the green light.

Or that often during busy times there's a policeman directing traffic in some intersections. Do autonomous cars know how to follow police hand signals, yet?

However, I understand there are drivers in the cars, so such things won't actually end up being a problem. Probably these are the kinds of reasons they want to test somewhere like Manhattan. Not because they've necessarily solved the problems, but because the only way to really solve them is by getting real-world data to work with.

True true, and good point about traffic police. The cars will need to be able to handle a popup traffic cop.

Also, what to do when a traffic light is just dead. A few weeks ago on 14th and 1st, the lights were just off. No traffic cop had taken over yet and power seemed to be completely out so no flashing red or yellow. So it was just drivers very carefully approaching the intersection and negotiating with other drivers.

I do suspect however that the day after these rollout, we'll have the answer to what happens in a game of chicken between a pedestrian and an autonomous vehicle.

How about this set of rules? It has fallback mechanism for such kinds of situations. As somebody from continental Europe, I sometimes wonder how wildly different and varied the world is.

 

[NuckFutter]

Although I heard India is just on another level entirely, but I can't speak from experience.

I've traveled in India several times, though only as a passenger. It is indeed on a whole other level. It's a pants-shitting experience the first several times until you realize that the drivers there have it under control.

The most important rule seems to be that there are no rules, so always expect the unexpected. It's like the worlds most congested ballet, and in some respects, it's beautiful.

 

[TenderBabyMeat]

I would like to just express how happy I am to hear that there will be autonomous vehicles in Manhattan. Not because I live there or have ever been there, mind you, but because this is the first real indication I have seen that the people developing these feel they are a viable option in a region with... weather!

All the testing regions commonly reported on up to this point have been in places with relatively simple weather patterns. Southern California, Arizona, Nevada... Predominately drier, some rain yes, but generally speaking simpler in that regard. But I live somewhere with a full spectrum of weather patterns that includes some rather impressive snowfall come winter. From what I can tell from a quick search, Manhattan does not get near the average snowfall that hits my area, but it is still a bit of hope for me to see that snow may not completely knock out the possibility of being able to use one someday.

But with my luck they won't give the green light to them until the week after I retire and no longer need to daily commute. But I can still dream.

 

[rbruster]

I disagree. Pennsylvania is often rated as the worst roads by truckers in the entire USA and the greater Pittsburgh is especially bad with all the aforementioned problems (well not so much double and triple parking) plus the added benefit of having a horribly thought out road system that often has no rhyme or reason as to why roads where designed the way they where. It's a reason why Uber is testing their self driving cars here[/quote]

Downtown Pittsburgh is like Calcutta, but with people instead of cattle.

 

[Ochre_face]

It's pretty exciting to see the progress being made in navigating traffic. But I suspect there'll be an even harder barrier to overcome for true level 5 autonomy, and that's the dozens of little interactions that we currently have with drivers, to get exactly where we need to be dropped off for instance.

Try explaining to a car that you need to look for the yellow gate with the fir trees because the main entrance is closed after five.

Any ideas on how they'll handle that in the future? Not this GM project with backup drivers, but true autonomous cars.

 

[compgeek89]

I'd suspect Manhattan would probably be easier. Especially in the gridded sections. The roads are pretty straightforward, and speed limits are relatively low at 25MPH.

The toughest challenge would probably be pedestrians, but pedestrians usually stay out of the car's way.

Have you ever been in Manhattan? Pedestrians cross where ever they want, generally reading their phones at the time. Cabs and trucks stop randomly and cut across 3 lanes of traffic and just making a right turn can be impossible unless you are willing to slowly nudge your way into the mass of humanity on the crosswalk.

Remind me to take the subway when this is going on.

 

[halse]

personally I prefer driving in Manhattan to doing so in Brooklyn

I don't see Manhattan as being that much more challenging than San Francisco where the Bolts seem to be doing fine as seen in this video
https://www.youtube.com/watch?v=Vfgjemwc9NU

some of the above posters are underestimating how well the autonomous Bolts, Waymo etc are doing these days

 

[CommentatorForNow]

personally I prefer driving in Manhattan to doing so in Brooklyn

I don't see Manhattan as being that much more challenging than San Francisco where the Bolts seem to be doing fine as seen in this video
https://www.youtube.com/watch?v=Vfgjemwc9NU

some of the above posters are underestimating how well the autonomous Bolts, Waymo etc are doing these days

That video for Cruise is really impressive. Of course it navigated the normal things well, but see how it managed:

(1) random jaywalker
(2) construction zone with construction workers
(3) double-parked truck causing single-lane (two direction) traffic

Etc. Etc.

But it's also impressive what it didn't do. It didn't pause to check its phone. It didn't spill its coffee on its lap. It didn't zone out at a stop light and forget to accelerate when the light changed... etc.

 

[Tom the Melaniephile]

I realize my opinion may not be popular, especially on a tech news site, but I just don't like where all this is going. It can be argued either way, but the side I am on is that this will ultimately dumb down our society. I refer to the movie "Wall-E" countless times these days as I think into what the future might look like. There will one day be a generation that has no idea how to drive a vehicle 'manually'. Think "Demolition Man"...

Just to put it in perspective, most Americans have no idea how to drive a horse. Nor how to safely crank-start a car.


And it really doesn't matter. Technology moves on. Hopefully we will stop having another 9/11 of deaths on our highways every month. Human drivers suck.

 

[Tom the Melaniephile]

However, as we can see with the coal industry in Appalachia, destroying the jobs is easy. Getting the people retrained and employed is a different issue altogether, especially the older the worker is.

Here ya go.

https://qz.com/990192/a-chinese-company ... d-farmers/

If the problem is jobs not the ideology of wanting to be a coal miner - wind and solar have far more jobs per $ invested than coal.

 

[artvandelayIIII]

I realize my opinion may not be popular, especially on a tech news site, but I just don't like where all this is going. It can be argued either way, but the side I am on is that this will ultimately dumb down our society.

As someone struck by a taxi in Manhattan, as well as having rid in a few taxis in Manhattan, I welcome the transition to autonomous cars. Driving a car simply requires a ton of concentration, quick reflexes, and calm emotional response, and no human is capable of being up to the task 100% of the time. There's simply too many distractions and variables. Even the best driver fails occasionally, although they often luck out when other drivers, pedestrians, etc. who do happen to be paying attention at that moment compensate.

Better to just take the task away from them. I typically use mass transit to commute to work, and I don't feel I've become dumber by removing myself from the driver's seat. Instead, I can use the commute time to actually pay attention to the music/podcast I'm listening to, or read, or play a game, or respond to emails... etc. Even talking on the phone (speakerphone) in a car has been shown to significantly impact driver reaction speeds.

While there's a few that actually just like the act of driving and give it the proper respect and attention, for the majority of people they'd rather be doing something else, and often try. The commute isn't a joy, but a chore. They're not being intellectually stimulated, but are becoming bored and distracted.

And all of this is without accounting for emotions. Automated cars aren't going to experience road rage, or start speeding, taking corners fast, etc. because they're running late. No rubbernecking at accidents on the road (which there should hopefully be less of). Or making errors about their state of wakefulness or intoxication.

People have only been driving cars up to this point because we're the only beings remotely capable of it. But objectively speaking, as a group, we're pretty terrible at it, and it's something that should be handed over to automation as soon as it's safe to do so.

As a fellow New Yorker and pedestrian, I 100% agree. I'm pretty much certain a robot taxi isn't going to make an aggressive turn into a crosswalk or blatantly run a red light toward a crowd of people trying to cross the street. I do have a feeling that NYC's finest human drivers are going to just cut off every self-driving car relentlessly and it'll be a very frustrating ride for anyone inside. But I could care less, the subway is faster much of the time and far cheaper to boot.

 

[artvandelayIIII]

I'd suspect Manhattan would probably be easier. Especially in the gridded sections. The roads are pretty straightforward, and speed limits are relatively low at 25MPH.

The toughest challenge would probably be pedestrians, but pedestrians usually stay out of the car's way.

Haha, never been to New York? I wanna see how this deals with specific intersections like anything along Times Square where you need to inch up until the point where the pedestrians decide to go behind you instead of in front of you. And only then are you able to actually go through the intersection.

Or the grid lock version where if you don't force your way into the intersection, the opposing traffic will eventually block you out again once they get the green light.

Or that often during busy times there's a policeman directing traffic in some intersections. Do autonomous cars know how to follow police hand signals, yet?

However, I understand there are drivers in the cars, so such things won't actually end up being a problem. Probably these are the kinds of reasons they want to test somewhere like Manhattan. Not because they've necessarily solved the problems, but because the only way to really solve them is by getting real-world data to work with.

True true, and good point about traffic police. The cars will need to be able to handle a popup traffic cop.

Also, what to do when a traffic light is just dead. A few weeks ago on 14th and 1st, the lights were just off. No traffic cop had taken over yet and power seemed to be completely out so no flashing red or yellow. So it was just drivers very carefully approaching the intersection and negotiating with other drivers.

I do suspect however that the day after these rollout, we'll have the answer to what happens in a game of chicken between a pedestrian and an autonomous vehicle.

If the cars don't stop for pedestrians in a crosswalk, they will need to be removed from our streets, no question, no room for negotiation. Frankly, there are a lot of part of the city where we shouldn't even HAVE cars at all.

 

[PlatinumSplBane]

As much as I love technological progression, (who honestly wants to go back to mid-90s tech?), I'm afraid that parts of society's progression will drag behind technological progression and we may experience some severe growing pains. Not enough jobs to go around coupled with more and more jobs being automated out of existence with a heavy dose of "You don't work, you don't eat, fuck you" mindset, which takes about two seconds to find examples of, would be pure pain and misery for countless number of people.That worries me.

I too, am no Luddite, but I can easily see that this kind of automation will cause societal havoc. Millions of jobs will be lost world wide, and this is just the beginning. I don't welcome this at all.

I'm not FOND of the idea, but your alarmist views fail to take into account a little something called "history".

Millions of jobs will be CREATED when you lose millions of jobs. The fact is no matter what industry you want to look at, once you get a paradigm shift in that industry (transportation going from horse to automobile, for example), jobs are ALWAYS lost. Blacksmiths and livery stables became non-viable due to vast oversupply of those services in the economy.

But a major increase in job availability came as a result of new road construction, blacksmiths became car mechanics, livery stables became gas stations and everyone had jobs again.

That's just ONE example of how an economy handles changes in technology. It's happened time and again when people simply find new vocations/occupations to handle changes in how things are done.

So the whole doom and gloom, "the sky is falling" and "millions will be unemployed" nonsense is, well, pure bullshit from an overall point of view. The jobs will be there in some form or another.

To be frank, you're speaking in generalities. Of course the workforce dynamic changes over a long period of time, but what happens in the short term... when you have many many more individuals, coupled with across the board automation in the service industry. Say 50,000 taxi/livery drivers are displaced in a given metro area. Will the majority of them find equivalent jobs, commensurate with their skillsets? Service jobs, probably with the exception of health care, are being displaced left and right, even in traditionally white collar "safe" jobs. Some IT professionals are being forced to reckon with the consequences of where technology is leading us, bravely and swiftly, that they are contemplating a "basic income" that everyone has to receive. For that to happen, at least in the US, we will need a shift in values.

 

[jimthomasUS]

The future can't come quick enough.

 

[Chuckstar]

I can think of literally thousands of labor categories. Your artificial restriction of what a labor category is is nonsensical, and bears no relationship to how automation affects the workforce.

Ok, maybe you could list some of these? And maybe explain to me why my analysis is wrong (other than "it just is"). It may well be. I'm not an expert. But I'm trying to see where I am wrong beyond that I "just am".

I defined a labor "category" to be what it is so that I could describe how the hierarchy of complex decision making works. If you don't like that term, that is fine. We can call it something else. Let's call it the "complex decision making hierarchy".

As automation works its way up this hierarchy, there is no corresponding increase in complexity at the top end that we humans can crowd into. What broad "classes of complex decision making" jobs are out there still? Let's say that mechanical engineering gets automated. What kinds of jobs would these engineers go into that isn't similarly complex and, therefore, also likely to be automated by the same technology? That's what I meant by "categories". Once a level of complex decision making is automated, it is automated across the board at that level and we humans have to crowd into what remains. So far there has been enough space for most of us (numerically speaking), but that space is simply going to shrink more and more. There is an upper limit to the complex decision making that a human can do, and we are engineering machines that are encroaching in on the bottom end of that realm already.

Labor categories:

Machinist
Nurse
Clerk
Cashier
Waiter
Cook
Programmer
Teacher
Truck driver
Warehouse picker
Lawyer
Accountant
Cinematographer
Painter
Cleric
Policeman


Etc.

Each of those will have different impact from automation. The idea that you can lump people into a category like “physical labor”, and then can make any useful statement about how automation will affect all workers in that set the same way is patently absurd.

 

[Chuckstar]

As much as I love technological progression, (who honestly wants to go back to mid-90s tech?), I'm afraid that parts of society's progression will drag behind technological progression and we may experience some severe growing pains. Not enough jobs to go around coupled with more and more jobs being automated out of existence with a heavy dose of "You don't work, you don't eat, fuck you" mindset, which takes about two seconds to find examples of, would be pure pain and misery for countless number of people.That worries me.

I too, am no Luddite, but I can easily see that this kind of automation will cause societal havoc. Millions of jobs will be lost world wide, and this is just the beginning. I don't welcome this at all.

I'm not FOND of the idea, but your alarmist views fail to take into account a little something called "history".

Millions of jobs will be CREATED when you lose millions of jobs. The fact is no matter what industry you want to look at, once you get a paradigm shift in that industry (transportation going from horse to automobile, for example), jobs are ALWAYS lost. Blacksmiths and livery stables became non-viable due to vast oversupply of those services in the economy.

But a major increase in job availability came as a result of new road construction, blacksmiths became car mechanics, livery stables became gas stations and everyone had jobs again.

That's just ONE example of how an economy handles changes in technology. It's happened time and again when people simply find new vocations/occupations to handle changes in how things are done.

So the whole doom and gloom, "the sky is falling" and "millions will be unemployed" nonsense is, well, pure bullshit from an overall point of view. The jobs will be there in some form or another.

To be frank, you're speaking in generalities. Of course the workforce dynamic changes over a long period of time, but what happens in the short term... when you have many many more individuals, coupled with across the board automation in the service industry. Say 50,000 taxi/livery drivers are displaced in a given metro area. Will the majority of them find equivalent jobs, commensurate with their skillsets? Service jobs, probably with the exception of health care, are being displaced left and right, even in traditionally white collar "safe" jobs. Some IT professionals are being forced to reckon with the consequences of where technology is leading us, bravely and swiftly, that they are contemplating a "basic income" that everyone has to receive. For that to happen, at least in the US, we will need a shift in values.

First of all, 50,000 taxi drivers aren’t losing there jobs all at once. This isn’t The Fifth Element

Second, what you’re describing has been happening across a variety of industries for generations. How many seamstresses do you know? 50-60 years ago, there were probably as many seamstresses in the U.S. as there are taxi drivers today.

 

[jeandemeung]

My worry would be software quality. How many lines of code, how many undiscovered features where it does exactly what the coder believed he had been told to do, but actually in these particular circumstances it has killed half a dozen bystanders.

The first thing to look at is the software assurance and management procedures, are they really up to the task? Have they applied rigorous methods to prove it? Ones commensurate with the risk to life that is involved?

I doubt it. It seems likely that to be safe enough for mass use the environment will have to be much more restricted and controlled, and that we will simply have to give up on mixing human and computer driven vehicles - not to mention the bikes and walkers.

They can still be a great boon. But the way this is being done the potential for unforeseen disasters is huge.

I'm don't share your concern there. We have tons of examples of life and death scenarios where we rely on software. Medical devices. Flying. Even existing cars with drive by wire.

Whether the self driving capabilities will be up to the standards we need for it to be released into the work remains to be seen. But rigorous testing is being performed, the above article describing one example of that. In time there will be enough miles on the system in enough environments that we can safely say that most of the bugs are worked out.

Eventually a bug might cause a fatality, but we have to accept that and understand that as we fix these bugs they are fixed for millions of cars at once (assuming we don't run them on android and they never get updated ).

The auto industry, and self driving cars in particular, does not use the same software procedures as the aircraft industry does. Not subject to the same audits. Testing as most on Ars will know is not a matter of throw it out into the environment and look for bugs as it behaves.

I agree that we depend on software all the time. And that fly by wire is real. And that, in aircraft, it works almost all the time, and in any case probably better as a whole than without it. But in the auto industry, and the home appliance industry, it works not all, but only most of the time, and sometimes it fails unexpectedly.

Some of the commenters here are right to say that self drive has the potential by eliminating human error and misconduct to dramatically lower traffic casualties, and that these are at a level which represents a human disaster. Six million died in the Holocaust. We kill double that every decade on the roads globally and no-one notices. Some here even defend it, and that is baffling. And that is not mentioning the indirect costs and deaths due to air pollution. Or the injuries, many crippling.

So we should applaud the arrival of self driving. But its not clear that we can or should get there by trying to make self driving cars work like very good people driven cars in the same defective by design environments which lead to the present wicked death and injury rates, and by mixing self drive with bikes, human driven vehicles and walkers in yet another element in this disastrous mix.

It seems likely that we will get there in the end by having to ban human drivers from routes used by the self driving. The quicker we get to the point where there are no human driven cars, the better, and if this means dedicating routes to self drive, the quicker we get to that, the better.

 

[Chuckstar]

I'd suspect Manhattan would probably be easier. Especially in the gridded sections. The roads are pretty straightforward, and speed limits are relatively low at 25MPH.

The toughest challenge would probably be pedestrians, but pedestrians usually stay out of the car's way.

Haha, never been to New York? I wanna see how this deals with specific intersections like anything along Times Square where you need to inch up until the point where the pedestrians decide to go behind you instead of in front of you. And only then are you able to actually go through the intersection.

Or the grid lock version where if you don't force your way into the intersection, the opposing traffic will eventually block you out again once they get the green light.

Or that often during busy times there's a policeman directing traffic in some intersections. Do autonomous cars know how to follow police hand signals, yet?

However, I understand there are drivers in the cars, so such things won't actually end up being a problem. Probably these are the kinds of reasons they want to test somewhere like Manhattan. Not because they've necessarily solved the problems, but because the only way to really solve them is by getting real-world data to work with.

True true, and good point about traffic police. The cars will need to be able to handle a popup traffic cop.

Also, what to do when a traffic light is just dead. A few weeks ago on 14th and 1st, the lights were just off. No traffic cop had taken over yet and power seemed to be completely out so no flashing red or yellow. So it was just drivers very carefully approaching the intersection and negotiating with other drivers.

I do suspect however that the day after these rollout, we'll have the answer to what happens in a game of chicken between a pedestrian and an autonomous vehicle.

If the cars don't stop for pedestrians in a crosswalk, they will need to be removed from our streets, no question, no room for negotiation. Frankly, there are a lot of part of the city where we shouldn't even HAVE cars at all.

What are you talking about? If one waits for the crosswalk to clear in Manhattan, without inching forward to bully one’s way between pedestrians, one can easily be stuck at the same intersection for 20-30 minutes. Even with that kind of attempt to inch through pedestrians, I’ve found myself stuck waiting for pedestrian for 2-3 minutes. Then like 1-2 cars get through, and the next cars wait 2-3 minutes more. That’s a recipe for gridlock if the cars can’t make their way through the sea of pedestrians.

 

[Chuckstar]

It's pretty exciting to see the progress being made in navigating traffic. But I suspect there'll be an even harder barrier to overcome for true level 5 autonomy, and that's the dozens of little interactions that we currently have with drivers, to get exactly where we need to be dropped off for instance.

Try explaining to a car that you need to look for the yellow gate with the fir trees because the main entrance is closed after five.

Any ideas on how they'll handle that in the future? Not this GM project with backup drivers, but true autonomous cars.

Smartphone app.

You’ll be able to set a location/route via overhead view map, or even “drive” the car by getting a feed from the forward camera. Of course, you won’t really be driving, in the latter case. You will be more pointing/guiding while the car retains the role of avoiding collisions — basically similar to being in the passenger seat telling the driver “now turn here”.

You’ll be able to authenticate others to control the car, like if a parking attendant or mechanic needs to be able to move the car around.

 

[novelski]

As much as I love technological progression, (who honestly wants to go back to mid-90s tech?), I'm afraid that parts of society's progression will drag behind technological progression and we may experience some severe growing pains. Not enough jobs to go around coupled with more and more jobs being automated out of existence with a heavy dose of "You don't work, you don't eat, fuck you" mindset, which takes about two seconds to find examples of, would be pure pain and misery for countless number of people.That worries me.

I too, am no Luddite, but I can easily see that this kind of automation will cause societal havoc. Millions of jobs will be lost world wide, and this is just the beginning. I don't welcome this at all.

I'm not FOND of the idea, but your alarmist views fail to take into account a little something called "history".

Millions of jobs will be CREATED when you lose millions of jobs. The fact is no matter what industry you want to look at, once you get a paradigm shift in that industry (transportation going from horse to automobile, for example), jobs are ALWAYS lost. Blacksmiths and livery stables became non-viable due to vast oversupply of those services in the economy.

But a major increase in job availability came as a result of new road construction, blacksmiths became car mechanics, livery stables became gas stations and everyone had jobs again.

That's just ONE example of how an economy handles changes in technology. It's happened time and again when people simply find new vocations/occupations to handle changes in how things are done.

So the whole doom and gloom, "the sky is falling" and "millions will be unemployed" nonsense is, well, pure bullshit from an overall point of view. The jobs will be there in some form or another.

To be frank, you're speaking in generalities. Of course the workforce dynamic changes over a long period of time, but what happens in the short term... when you have many many more individuals, coupled with across the board automation in the service industry. Say 50,000 taxi/livery drivers are displaced in a given metro area. Will the majority of them find equivalent jobs, commensurate with their skillsets? Service jobs, probably with the exception of health care, are being displaced left and right, even in traditionally white collar "safe" jobs. Some IT professionals are being forced to reckon with the consequences of where technology is leading us, bravely and swiftly, that they are contemplating a "basic income" that everyone has to receive. For that to happen, at least in the US, we will need a shift in values.

First of all, 50,000 taxi drivers aren’t losing there jobs all at once. This isn’t The Fifth Element

Second, what you’re describing has been happening across a variety of industries for generations. How many seamstresses do you know? 50-60 years ago, there were probably as many seamstresses in the U.S. as there are taxi drivers today.

What timeframe is all at once? In a year or 10 years? Most likely there won't be such a swift change that will make drivers unemployed in a year. But period of 10 years is something different and depending on age structure of people employed, it might be a problem. Especially considering that technology for self-driving cars (e.g. object recognition) can be used in other disciplines and that can have effects on other jobs available.

But the whole argument of automation destroying jobs is not about sudden disappearing of entire job categories, but about sufficiently high structural unemployment. And not only of unskilled workforce, but also specialists and educated people. Say automation makes every 5th job position redundant. That's ~20% unemployment that needs to be somehow assimilated into known jobs, or some completely new jobs have to emerge. So far, increased consumption and more services have made those jobs, but every person can own only finite amount of things and spend finite amount of time on services.

 

[haar]

“Still, as Frank might have sung, an AV that can make it there can make it anywhere...”

and “how do you get to...” ... practice, practice, practice...

what happens when to car is low on gas?... “i am sorry, Dave, i can not do that... “


btw, the car should excel at this... it is all about awareness... people running out, slow traffic. all about patience... does a self-driving car have a method to “flip the bird” or give the finger... LOL

 

[andygates]

"Maybe it'll work on the test track, but it'll never work in real life!"

"Okay, maybe it'll work at low speeds in suburbia, but it'll never work in real life!"

"Okay, maybe it'll work at regular speeds in regular weather, but it'll never work in real life!"

"Okay, maybe it'll work at in the Valley, but it'll never work in real life!"

...now Manhattan. Naysayers blind to the rather obvious incremental progress are getting backed into a snowdrift in Mumbai as their remaining edge cases.

 

[compgeek89]

"Maybe it'll work on the test track, but it'll never work in real life!"

"Okay, maybe it'll work at low speeds in suburbia, but it'll never work in real life!"

"Okay, maybe it'll work at regular speeds in regular weather, but it'll never work in real life!"

"Okay, maybe it'll work at in the Valley, but it'll never work in real life!"

...now Manhattan. Naysayers blind to the rather obvious incremental progress are getting backed into a snowdrift in Mumbai as their remaining edge cases.

The only reason I am more skeptical on Manhattan is because the only way to make any kind of forward progress in some parts of town is to do things that are not quite legal... E. G. Inching into the crosswalk on a turn, driving too close to the car in front to avoid someone darting in, being partly in the far side of the intersection when the light turns red. Can you see any programmer putting this into the car programming?

To be clear, I agree with the poster above... the cars will be perfectly safe, but a ride crosstown will take 2 hours. They will be about as efficient as the horse drawn carriages. As long as they don't have too many of them, there will be no issue. But if a whole load are on the streets, look for some fun gridlock.

 

[nbs2]

I'd suspect Manhattan would probably be easier. Especially in the gridded sections. The roads are pretty straightforward, and speed limits are relatively low at 25MPH.

The toughest challenge would probably be pedestrians, but pedestrians usually stay out of the car's way.

I guess you've not been to NYC?

It's not that NYC is particularly unique in the types of challenges presented. SF and DC for example also have lots of pedestrians, bikes and bike courriers, utility vehicles, roadwork, double/triple parking, messy intersections, etc.

But NYC might bring these things to the next level. Everyone just does whatever they want, ranging from other drivers to pedestrians. Jaywalking is practically the standard, and people totally assume that cars will stop for them. People just pull over randomly without blinkers, block side streets all the time, and speed limits and red lights are often taken more as suggestions than hard rules.

I drive in DC daily and drove in SF several times, but subjectively, I feel like NYC is a bit more challenging to deal with. In my line of work, I travel a fair amount and have driven in some pretty crazy places (eg. Haiti where there are no rules, lights, speed limits etc.) and I've always felt like driving in NYC and DC has prepped me very well for these situations. Although I heard India is just on another level entirely, but I can't speak from experience.

I've visited and driven on the streets of NYC and her aforementioned gridlock. I lived and drove the streets and highways of DC and her sprawling jurisdictional transitions. I live and drive on the highways of Houston and her simultaneous 11-lane jams and 95 mph open lanes.

And I've ridden in cars through many of the metropolises and major cities of India - I'm not sure you could pay me enough to actually drive there. There is a reason tuk-tuks/three-wheelers and their lawnmower engines are viable forms of transport.

 

[KGFish]

As much as I love technological progression, (who honestly wants to go back to mid-90s tech?), I'm afraid that parts of society's progression will drag behind technological progression and we may experience some severe growing pains. Not enough jobs to go around coupled with more and more jobs being automated out of existence with a heavy dose of "You don't work, you don't eat, fuck you" mindset, which takes about two seconds to find examples of, would be pure pain and misery for countless number of people.That worries me.

There is no reason to think automation reduces employment on a national scale. It changes what jobs there are and what they pay. It increases productivity, and improves quality in the industries where it is applied. But there is no evidence that on a national scale it leads to unemployment. And there is no reason to think that banning it or freezing it will lead, nationally, to better outcomes.

It may be that if you are an old fashioned print operator, the move to electronic typesetting lowers your premium and your wages. But there is no reason to think that the newspaper industry moving to electronics lowers employment for the whole country. In fact, it probably increases it. But it does lower the number of typesetters.

However, as we can see with the coal industry in Appalachia, destroying the jobs is easy. Getting the people retrained and employed is a different issue altogether, especially the older the worker is.

Especially since frequently, the new jobs aren't where the old jobs are. And moving is an incredibly expensive undertaking, both financially and emotionally.

 

[veldrin]

The bigger challenges for AVs are things like the article mentioned, primarily negotiating complicated scenarios where the conservative programming may cause the AV to simply be incapable of proceeding (e.g. where pedestrians refuse to wait for the AV to proceed.

On the flip side, an AV can safely deploy techniques that human drivers cannot (as well, anyway). An AV can slowly creep through a crowd in a way that slow human reaction times can only accommodate with the active acquiescence of the pedestrians. Basically do what cab drivers often try to do, but without occasionally striking pedestrians as often happens now.

As time goes on, I suspect we will revert to much less control at urban intersections since the cars will communicate with each other to figure out the most efficient way to move them through and the cars will just avoid the pedestrians as necessary.

 

[afidel]

My worry would be software quality. How many lines of code, how many undiscovered features where it does exactly what the coder believed he had been told to do, but actually in these particular circumstances it has killed half a dozen bystanders.

The first thing to look at is the software assurance and management procedures, are they really up to the task? Have they applied rigorous methods to prove it? Ones commensurate with the risk to life that is involved?

I doubt it. It seems likely that to be safe enough for mass use the environment will have to be much more restricted and controlled, and that we will simply have to give up on mixing human and computer driven vehicles - not to mention the bikes and walkers.

They can still be a great boon. But the way this is being done the potential for unforeseen disasters is huge.

I'm don't share your concern there. We have tons of examples of life and death scenarios where we rely on software. Medical devices. Flying. Even existing cars with drive by wire.

Whether the self driving capabilities will be up to the standards we need for it to be released into the work remains to be seen. But rigorous testing is being performed, the above article describing one example of that. In time there will be enough miles on the system in enough environments that we can safely say that most of the bugs are worked out.

Eventually a bug might cause a fatality, but we have to accept that and understand that as we fix these bugs they are fixed for millions of cars at once (assuming we don't run them on android and they never get updated ).

The auto industry, and self driving cars in particular, does not use the same software procedures as the aircraft industry does. Not subject to the same audits. Testing as most on Ars will know is not a matter of throw it out into the environment and look for bugs as it behaves.

I agree that we depend on software all the time. And that fly by wire is real. And that, in aircraft, it works almost all the time, and in any case probably better as a whole than without it. But in the auto industry, and the home appliance industry, it works not all, but only most of the time, and sometimes it fails unexpectedly.

Huh? ISO 26262 (coding standards for automotive programming) and its parent IEC 61508 are very similar to DO-178C (coding standards for the avionics industry).