Forum adverts like this one are shown to any user who is not logged in. Join us by filling out a tiny 3 field form and you will get your own, free, dakka user account which gives a good range of benefits to you: No adverts like this in the forums anymore. Times and dates in your local timezone. Full tracking of what you have read so you can skip to your first unread post, easily see what has changed since you last logged in, and easily see what is new at a glance. Email notifications for threads you want to watch closely. Being a part of the oldest wargaming community on the net. If you are already a member then feel free to login now.

 Kilkrazy wrote:
http://www.bbc.co.uk/news/world-us-canada-43604440

Tesla car involved in fatal crash while on auto-pilot.

Thats a... wildly inaccurate summary.
More accurate would be 'Tesla car with driver assist (unfortunately named the Autopilot system) failed to bend the laws of physics in under five seconds, after driver decided to point the car at the concrete barrier and take his hands off the wheel'

Just Tony wrote:I'm starting to really like the concept of the hive city, and having layers of street so as to have traffic be slowed less. TONS of infrastructure needed to make that dream a reality, but if they can pull in off in some place like Manhattan, then we're looking at a viable alternative.

Interesting question: who came up with the idea of the hive city? Asimov used to concept in Caves of Steel, but I don't know where else it was used, or possibly before that.

E M Forster's "The Machine Stops" (1908) mentions hive habitation?

Voss wrote:

 Kilkrazy wrote:
http://www.bbc.co.uk/news/world-us-canada-43604440

Tesla car involved in fatal crash while on auto-pilot.

Thats a... wildly inaccurate summary.
More accurate would be 'Tesla car with driver assist (unfortunately named the Autopilot system) failed to bend the laws of physics in under five seconds, after driver decided to point the car at the concrete barrier and take his hands off the wheel'

Clickbait title is clickbait.

Sure, Autopilot as a name sounds more autonomous than a "driving assistance" feature, but the driver should have known this. Taking your hands off the wheel for 6 seconds and driving straight into a concrete barrier is hardly the fault of the driving assistance.

Thing is was the driver trained in how the assisted driving features would work or was it a case of "here's a car, there's a manual inside off you go".

This harkens back to my view that a lot of the issues with driving are linked back to a lack of sufficient training and too much relying on the idea that drivers will read and understand the manual.

The driver might well have figured that the car would see the barrier, make the turn itself and correct the situation -- esp if the car has some assisted features where it will turn the car itself whilst driving to aid the driver. Might even be the driver knew they were driving into the barrier at the wrong angle; decided that they couldn't correct the action in that time and that instead they'd let the faster correction features of the car correct the issue;

Voss wrote:

 Kilkrazy wrote:
http://www.bbc.co.uk/news/world-us-canada-43604440

Tesla car involved in fatal crash while on auto-pilot.

Thats a... wildly inaccurate summary.
More accurate would be 'Tesla car with driver assist (unfortunately named the Autopilot system) failed to bend the laws of physics in under five seconds, after driver decided to point the car at the concrete barrier and take his hands off the wheel'

It's not a summary, it's a headline.

from/in a Holiday Inn, apparently.

https://twitter.com/dancanon/status/979209388937437184

https://twitter.com/tom_peters/status/981497202957803522

Work to do. Current issue of Economist, special section on AI. One European bank has asked Infosys to help it cut operations staff from 50,000 to 500. (No typo.)

 reds8n wrote:
from/in a Holiday Inn, apparently.

https://twitter.com/dancanon/status/979209388937437184

What's going on? I can't view the video.

 AndrewGPaul wrote:

 reds8n wrote:
from/in a Holiday Inn, apparently.

https://twitter.com/dancanon/status/979209388937437184

What's going on? I can't view the video.

Room service being delivered by a robot.

Pretty cool I'd say.

What I'm noticing is that it's the service industry that's getting hit with these things now. Manufacturing isn't automating NEARLY as fast. So your McJobs are going way, but your GMJobs are not going away, at least not in the same vein.

 Just Tony wrote:
What I'm noticing is that it's the service industry that's getting hit with these things now. Manufacturing isn't automating NEARLY as fast. So your McJobs are going way, but your GMJobs are not going away, at least not in the same vein.

Manufacturing isn’t automating as fast right now because manufacturing has been automating for decades now. Manufacturing already employs far fewer workers than it used to due to the persistent increase in automating assembly lines. The service sector has been automating much slower because there is a much lower skill threshold to get into the service sector labor pool so labor costs wee lower so the service sector needed to wait for the cost of automation to go down/technology to improve. Now that we’re at that point we are seeing more service sector/food industry jobs get replaced with automation.

Prestor Jon wrote:

 Just Tony wrote:
What I'm noticing is that it's the service industry that's getting hit with these things now. Manufacturing isn't automating NEARLY as fast. So your McJobs are going way, but your GMJobs are not going away, at least not in the same vein.

Manufacturing isn’t automating as fast right now because manufacturing has been automating for decades now. Manufacturing already employs far fewer workers than it used to due to the persistent increase in automating assembly lines. The service sector has been automating much slower because there is a much lower skill threshold to get into the service sector labor pool so labor costs wee lower so the service sector needed to wait for the cost of automation to go down/technology to improve. Now that we’re at that point we are seeing more service sector/food industry jobs get replaced with automation.

There was a word you breezed past in my post: now. I said that the service industry is getting hit with automation now, as in that's the focus. Since I've been working in industry for over two decades, I know EXACTLY what the rate of automation is, and what jobs can't be automated.

 Just Tony wrote:

Prestor Jon wrote:

 Just Tony wrote:
What I'm noticing is that it's the service industry that's getting hit with these things now. Manufacturing isn't automating NEARLY as fast. So your McJobs are going way, but your GMJobs are not going away, at least not in the same vein.

Manufacturing isn’t automating as fast right now because manufacturing has been automating for decades now. Manufacturing already employs far fewer workers than it used to due to the persistent increase in automating assembly lines. The service sector has been automating much slower because there is a much lower skill threshold to get into the service sector labor pool so labor costs wee lower so the service sector needed to wait for the cost of automation to go down/technology to improve. Now that we’re at that point we are seeing more service sector/food industry jobs get replaced with automation.

There was a word you breezed past in my post: now. I said that the service industry is getting hit with automation now, as in that's the focus. Since I've been working in industry for over two decades, I know EXACTLY what the rate of automation is, and what jobs can't be automated.

I'm not trying to 'gotcha' here, but i do find it amusing that you're willing to bet the speed of light can be broken by as-yet-unheard of means, but that a fully automated manufacturing plant is beyond our capability

Speaking of service industry robots: Self checkouts. These fething things are a pain. Unless the store is practically deserted, it always faster to queue up and get checked out by a human.

 feeder wrote:

 Just Tony wrote:

Prestor Jon wrote:

 Just Tony wrote:
What I'm noticing is that it's the service industry that's getting hit with these things now. Manufacturing isn't automating NEARLY as fast. So your McJobs are going way, but your GMJobs are not going away, at least not in the same vein.

Manufacturing isn’t automating as fast right now because manufacturing has been automating for decades now. Manufacturing already employs far fewer workers than it used to due to the persistent increase in automating assembly lines. The service sector has been automating much slower because there is a much lower skill threshold to get into the service sector labor pool so labor costs wee lower so the service sector needed to wait for the cost of automation to go down/technology to improve. Now that we’re at that point we are seeing more service sector/food industry jobs get replaced with automation.

There was a word you breezed past in my post: now. I said that the service industry is getting hit with automation now, as in that's the focus. Since I've been working in industry for over two decades, I know EXACTLY what the rate of automation is, and what jobs can't be automated.

I'm not trying to 'gotcha' here, but i do find it amusing that you're willing to bet the speed of light can be broken by as-yet-unheard of means, but that a fully automated manufacturing plant is beyond our capability

Speaking of service industry robots: Self checkouts. These fething things are a pain. Unless the store is practically deserted, it always faster to queue up and get checked out by a human.

You don't like the discount line?

Prestor Jon wrote:

Speaking of service industry robots: Self checkouts. These fething things are a pain. Unless the store is practically deserted, it always faster to queue up and get checked out by a human.

You don't like the discount line?

 feeder wrote:

 Just Tony wrote:

Prestor Jon wrote:

 Just Tony wrote:
What I'm noticing is that it's the service industry that's getting hit with these things now. Manufacturing isn't automating NEARLY as fast. So your McJobs are going way, but your GMJobs are not going away, at least not in the same vein.

Manufacturing isn’t automating as fast right now because manufacturing has been automating for decades now. Manufacturing already employs far fewer workers than it used to due to the persistent increase in automating assembly lines. The service sector has been automating much slower because there is a much lower skill threshold to get into the service sector labor pool so labor costs wee lower so the service sector needed to wait for the cost of automation to go down/technology to improve. Now that we’re at that point we are seeing more service sector/food industry jobs get replaced with automation.

There was a word you breezed past in my post: now. I said that the service industry is getting hit with automation now, as in that's the focus. Since I've been working in industry for over two decades, I know EXACTLY what the rate of automation is, and what jobs can't be automated.

I'm not trying to 'gotcha' here, but i do find it amusing that you're willing to bet the speed of light can be broken by as-yet-unheard of means, but that a fully automated manufacturing plant is beyond our capability

Speaking of service industry robots: Self checkouts. These fething things are a pain. Unless the store is practically deserted, it always faster to queue up and get checked out by a human.

I prefer to be paid for my labor, so I avoid the self-checkout.

The last time I went to the USA and hired a car, the gas stations had two lanes of pumps with different prices. The more expensive price was for human service, in other words, if you drove in there the man would come out and fill up your car for you. This was nearly 25 years ago, though. Do they still have them now?

 Kilkrazy wrote:
The last time I went to the USA and hired a car, the gas stations had two lanes of pumps with different prices. The more expensive price was for human service, in other words, if you drove in there the man would come out and fill up your car for you. This was nearly 25 years ago, though. Do they still have them now?

Depends. In most places that doesn't exist at all, in a few states it's still illegal to pump your own gas because the state doesn't want to allow those obsolete jobs to disappear.

The thing I notice with a lot of service sector automation is that its not adaptive automation. As a result if anyone wants to do something or request something "off the system" then they've got to flag down and wait for an attendant anyway.

Even if its just to casually discuss pros and cons of a purchase choice; and most would expect to get info from the attendant beyond what's on the product label (which many times does not hold "all" the information and also won't go into details such as ease of use, practical applications etc...).


I would also say that part of the service sector is salesmanship - ergo the art of encouraging a customer to part with their money. Put automation in and its great for those who already know what they want, but its not so great at those who don't know. Or (for the shop) those who can afford better but are looking at the lower end - that means less potential sales than a good salesman.

Then again a lot of, at least retail, jobs are not based on skilled salesmen anymore since they often cycle staff very regularly and keep a lot of staff on short term limited hours contracts - so many can serve well but have no incentive to learn the trade of what they sell.

An interesting tech report from the BBC, touching on some of the themes involved in this thread and in the Uber thread.

http://www.bbc.co.uk/news/technology-42516066

I've necroed this thread because there is new information about the case.

https://www.theguardian.com/technology/2018/may/08/ubers-self-driving-car-saw-the-pedestrian-but-didnt-swerve-report

TL/DR: Uber self-driving cars have software that let's them ignore false positive readings like plastic bags. The car read the pedestrian as a false positive, so it did not slow down for her.

On the plus side I'm willing to bet there's going to be some sensor interpretation redundancies now to give "oversight". I realize human judgment is fallible as well, but I can't help but think that AI judgment calls are going to get a black eye over this.

The thing to bear in mind is that AI "judgement calls" are based on the algorithms and data programmed into them by human engineers.

(Except in the case of machine learning where the AI network is allowed to learn by trial and error.)

What I'm getting at is that either we need to have people anticipate more potential logic bombs like this. If this is going to have a chance of working, it needs to basically be human error causing any fatalities. The industry can't afford something as catastrophic as "the AI chose to ignore it", which is how you KNOW it will be spun, especially by the transportation industry.

So essentially we'll have to put L plates on every single driver-free car because the driver will have to act like a driving instructor. Watching and monitoring and taking over from the car when required (after a training course in learning how the car thinks so that they can anticipate when the car will fail).

Granted if everyone drove as if everyone else were a beginner the roads might well be a lot safer.

the risk will be complacency. Everyone will be really attentive and observant for the first few weeks of owning the driverless car; but if the car never makes a failure on any trip out there's a high chance the person will stop paying attention. It takes a lot of concentration to pay attention to nothing and not allow your mind to wander - to turn your head to chat to the person next to you; to listen to that favourite music track; to glance at the scenery; to take a yawn and 50seconds nap because your'e super tiered but its ok the car is doing the work.





All that said cars could harvest vast amounts of data if regular cars are fitted with all the sensors the computer requires and then that data is fed back into a central archive - reviewed and added to the cars brain. Many newer edition cars already have cameras and sensors so adding in the option to opt into having this data go back to home-base for review would be a powerful teaching tool for the AI



Thinking on it more it wouldn't surprise me if, barring companies like Uber, we never actually see a marketed driverless car. Instead we'll see a steady increase in the number of machine driver aids. There's already good braking aids based on sensors which can be far faster to react than the driver. So one could expect to see more and more driver aids steadily creep out into the market until the majority of cars are heavily aid reliant.
Groups like Uber might push for the whole package, but I see the manufacture side being a steady advance. Policies like the UK governments push for all electric by 2020 might aid this as it forces the market to evolve and advance (although honestly I think that policy is a pie-in-the-sky dream - if not just from a production side, but also from a consumer side - not everyone can afford brand new cars all the time nor want to sign up to hire-purchase deals

 Just Tony wrote:
What I'm getting at is that either we need to have people anticipate more potential logic bombs like this. If this is going to have a chance of working, it needs to basically be human error causing any fatalities. The industry can't afford something as catastrophic as "the AI chose to ignore it", which is how you KNOW it will be spun, especially by the transportation industry.

We've never demanded zero fatalities from anything ever before. Right now we tolerate hundreds of thousands of car fatalities and we barely even talk about it. In fact, we just shrug and accept a long history of car manufacturers fighting against regs that made cars safer, and it never stopped us buying and driving cars.

As long as AI cars show a record of being clearly safer than human drivers they'll be adopted just fine.

 sebster wrote:
As long as AI cars show a record of being clearly safer than human drivers they'll be adopted just fine.

I kind of disagree with that. From an engineering point of view it's absolutely true. Zero fatalities is a stupid goal, and the moment automated cars can demonstrate a better safety record than human drivers they should be put into mass production. The system should continue to be refined, of course, but holding back on a major safety improvement because it isn't quite perfect yet is insanity. But we aren't dealing with a pure engineering decision. Automated cars are going to be a political decision, and public perception of the situation is clearly not rational. People will get stupid and emotional and blow every accident out of proportion, and politicians will listen to their idiot base and impose regulations on automated cars.

In contrast politicians/car companies could seriously push the news on car fatalities. Right now its accepted because its basically repressed news - or rather its not shouted about. There's the "serious crash" every so often, but there's no big news on how many are killed each year nor each day.

If it got political as much as the kills from automated cars can be bumped up the kills from human controlled ones can be blown up even more into a huge debate. Indeed its almost crazy to think how we are super strict with industries as to health and safety; and yet the most dangerous thing for most people is just travelling to work.

Automated cars will have to be preceded by a walking robot carrying a red warning flag.

 Kilkrazy wrote:
Automated cars will have to be preceded by a walking robot carrying a red warning flag.

Hey it only took over 2 decades full of accidents and fatalities to get Detroit to pioneer the kind of traffic laws and roadways we have today. Automated cars probably won’t take more than 12-15 years to sort out.

Since all this comes from a Tesla issue and going backup to the issue of automation it's worth noting that Tesla themselves had to roll back on automation because it was slowing down production.

Elon Musk admits robots are slowing down Tesla production – and says humans are the answer
https://www.telegraph.co.uk/news/2018/04/16/elon-musk-says-excessive-automation-slowed-production-tesla/

And that was a mistake already repeated from GM efforts a few decades back. At least in the automotive industry the flexibility and adaptability of humans still beat robots