Umm, how many lines of code do you think go into a rocket launch? The space shuttle had about 2 BILLION lines of code that all had to work approximately perfectly. Any rocket launched today, especially one that plans to remain reusable, is going to have a LOT of code made to a very high standard of reliability.
The space shuttle had a lot of code and NASA paid a huge amount each year to enhance and maintain it. It worked very reliably and the few bugs that were found never affected the shuttle. I have not seen any of the companies stepping up to that level and spending the same amount in dollars per line of code.
And if you think physics and engineering don't play a big role in self driving cars you don't understand the problem adequately.
Physics and engineering does play a major role in self-driving cars. The car has to have a physical model of itself (does that make it self-aware) in order to know how inputs to the car's systems, brakes, engine, steering will effect the vehicle. I am just saying that working out the physics and engineering problems are going to be easier than making sure all the code is working correctly.
Now it might be fair to say that launching a rocket is easier but it certainly isn't because of the number of lines of code. The challenge with autonomous vehicles isn't doing reliable code but in figuring out exactly what the code should do. Provided a company is willing to spend the money to do it, we know how to make reliable code. (we're just not used to companies actually going to the trouble) We haven't figured out how to design software that can recognize and react sanely to all the various inputs that occur in real world driving situations. The algorithms are just super difficult to figure out. The progress that has been made has been remarkable but the corner cases still unsolved remain fairly intimidating.
I agree with you. Figuring out exactly what to do, in all situations, and then responding appropriately is a very challenging task. The task is to make sure all the corner cases are covered. The next task is to develop code that covers all those situations, corner cases and responses without any defects being introduced that would harm someone.
I think there are times that AI will do better than humans. One of them is braking. A computer can detect when one of the tires is locking up. It can then decrease brake pressure to just that wheel. And it can do this hundreds of times a second. Another is detecting when a car is starting to hydroplane. Another time is when a car starts to skid. Onboard accelerometers could detect differences between steering input and change in car direction. And if there is a mismatch, take corrective action. The challenge is to marry them all into a system that can replace what humans do all so well.
And I am not expecting the AI to be perfect but I do expect it to be better than what it is replacing. Just eliminating the drunk drivers on the road would halve the accident fatalities. That in itself would be a good thing.
I wanted to illustrate situations that humans deal with all the time but can be difficult for computers. Can we train AI to handle adverse conditions, unexpected obstructs, and untimely situation? Sure we can. But it take time and data, with for unlikely events is scarce.
I worked in ML field. I know the limitations of it. Take Computer Vision, it didn't seem to help Uber's self-driving car that ran over and killed that person. Both radar and lidar are great but have their limits. What if they feed conflicting data to the computer? How does it decide?
I never implied this was on the level of biz logic on a web framework. I have coded some of the most sophisticated ML code and know that it still not ready to drive my car.
Could you code a computer to avoid a child running in the road?
Could you code a computer to stay in it's lane during a blizzard when the lane lines are obscured?
Could you code a computer to steer out of a skid on black ice?
Could you code a computer to drive in pea soup fog when the sensors are blinded?
The difference between human drivers and self-driving cars is hundreds of millions of lines of code that all have to work perfectly.
Hundreds of millions of lines of code? My brain does it like this:
f(current situation) -> best action
AI is the same.
That's one line of code.
The problem is:
1) deciding what f to use, which is an optimization problem. Easy.
2) what data comprises the "current situation" (cameras, LIDAR, etc.). Hard, and prone to human bias.
3) what do you mean by best (minimize accidents, minimize accidents involving casualties, minimize number of deaths, etc.?). Very Hard.
When you can define 2 and 3 well enough, 1 is easy. That is why AI can be useful.
SFD
One line of code except for the library call.
"f(current situation)" is a couple hundred million lines right there.
"best action" is another couple hundred million lines.
Don't forget the OS. You are not going to run this on Windows. You don't want the blue screen of death right as you are heading for a sharp curve.
The complexity of autonomous vehicles is immense, especially since the general public and regulators are expecting them to be better at making decisions and safer than human drivers. I'd be willing to say that it's orders of magnitude bigger than the difference between reusable and non-reusable rockets.
The difference between reusable and non-reusable rockets is physics and engineering.
The difference between human drivers and self-driving cars is hundreds of millions of lines of code that all have to work perfectly.
I don't know if these "You Tube Moderators" have looked around much, but you can find hateful comments on pretty much any video. Why block comments on just this one?
If YouTube wants to block "hateful comments" why do they allow comments anywhere?
Because the quantity and hatefulness blew the scale?
The ethics board for AI folds because of unethical behavior of the non-artificial intelligent types.
It seems that some at Google think that having an opinion different than their group-think is reason enough to try and silent them.
No tolerance for diversity of thought. No tolerance for different opinions. Hate them because they are different. That is their moral compass?
If I answered every email, I would not get any other work done. Most people cc the whole team for every email, creating huge email chains.
If you are answering all your emails then you are not doing any real work, AKA you are a manager.
This is the digital age's equivalent to a paper pusher.
The reply was meant to be tongue-in-cheek to the argument that the "market economy" and "competition" would push consumer to one company's product vs another's because of the carbon tax.
The new gizmos have to competitive economically or very nearly so for that carbon tax to have that effect. There is promising products and research that do offer lower carbon emissions but not all are competitive costwise with the products they are trying to replace.
That doesn't mean we shouldn't keep trying to reduce our carbon footprint where we can, just that the "green new deal" proposal that all these technologies will magically be ready in a decade is absurd.
If what you say is true then solar power plants would be cheaper than coal power plants, electric cars would be cheaper that gasoline cars, and electric planes would be cheaper to fly than jet aircraft. None of these green technologies have any carbon tax applied to them hence their products will be less expensive. Consumers will flock to them because of the lower prices and carbon emissions will plummet.
Apple could order a million of those tiny iPhone screws and it wouldn't take up more than a shelf in space. It could dual source them so it never runs out. One manufacturer in the US and one in China.
This is an easy solvable problem in supply chain management, something that you think Apple would be good at after 40+ years in business. That what happens when you outsource everything. It seems they outsource their business intelligence too. Now their brain-dead management can't figure it out.
Simple, IBM wants cheap skilled employees. Older skilled employees are expensive.
IBM is never going to fix it's skills gap until they value their entire workforce.
Who would want to work there when you are just a cog until you get older or the "hot technology of the day" changes and then they "resource action" you (IBM's term for layoffs).
Does your doctor work 100 hour weeks, how about your lawyer, the engineer designing the bridge you will be driving over, or the high rise you will be working in, or the pilot flying the plane you are traveling in? Ask the surgeon if he as performed 99 hours of surgery before he cuts you open for a 'simple one hour surgery. Or the pilot of a plane if he is putting in 100 hours of flying that week?
In all cases, the answer will be no but probably not even close to 100 hrs. But this is expected of software engineers, especially in the gaming industry. Treat them like digital garment workers. Wringing every last ounce out of them. Why should management care about their health, employee burnout, or work-life balance? If you don't like it, they say, "Just quit". They will just find the next naive inexperienced developer to fill the role.
There really is a simple solution to this all. Pay all software engineers by the hour. Overtime paid at time and a half. When the dev costs suddenly triple, management will take notice and stop this insane practice.
There certainly have been times in history where scientists have disputed with each other in less than sociable ways. Newton and Hooke, Newton and Leibniz, Chandrasekhar and Eddington to name a few. There have even been cases where a more prominent scientist has denied admission to the Royal Society of a scientific rival.
He is wrong, "physics was invented and built by physicists." But he was right, "it's not by invitation". It is not a social club. You don't get a invitation in the mail. You join by achievement, by accomplishment. All this gender talk is a distraction from real physics.
Accenture is coaching hundreds of Cobol programmers every year in India and the Philippines to work at banks.
So, if you are in the U.S. and you know Cobol already, you might get a few years of employment out of it. However, such jobs will go overseas, too.
You will have a job for a few years and then they will outsource your position. A few years later you can come back as a consultant to fix the spaghetti code that the overseas 'coders' ran through a blender.
Slashdot Mirror
Umm, how many lines of code do you think go into a rocket launch? The space shuttle had about 2 BILLION lines of code that all had to work approximately perfectly. Any rocket launched today, especially one that plans to remain reusable, is going to have a LOT of code made to a very high standard of reliability.
The space shuttle had a lot of code and NASA paid a huge amount each year to enhance and maintain it. It worked very reliably and the few bugs that were found never affected the shuttle. I have not seen any of the companies stepping up to that level and spending the same amount in dollars per line of code.
And if you think physics and engineering don't play a big role in self driving cars you don't understand the problem adequately.
Physics and engineering does play a major role in self-driving cars. The car has to have a physical model of itself (does that make it self-aware) in order to know how inputs to the car's systems, brakes, engine, steering will effect the vehicle. I am just saying that working out the physics and engineering problems are going to be easier than making sure all the code is working correctly.
Now it might be fair to say that launching a rocket is easier but it certainly isn't because of the number of lines of code. The challenge with autonomous vehicles isn't doing reliable code but in figuring out exactly what the code should do. Provided a company is willing to spend the money to do it, we know how to make reliable code. (we're just not used to companies actually going to the trouble) We haven't figured out how to design software that can recognize and react sanely to all the various inputs that occur in real world driving situations. The algorithms are just super difficult to figure out. The progress that has been made has been remarkable but the corner cases still unsolved remain fairly intimidating.
I agree with you. Figuring out exactly what to do, in all situations, and then responding appropriately is a very challenging task. The task is to make sure all the corner cases are covered. The next task is to develop code that covers all those situations, corner cases and responses without any defects being introduced that would harm someone.
I think there are times that AI will do better than humans. One of them is braking. A computer can detect when one of the tires is locking up. It can then decrease brake pressure to just that wheel. And it can do this hundreds of times a second. Another is detecting when a car is starting to hydroplane. Another time is when a car starts to skid. Onboard accelerometers could detect differences between steering input and change in car direction. And if there is a mismatch, take corrective action. The challenge is to marry them all into a system that can replace what humans do all so well.
And I am not expecting the AI to be perfect but I do expect it to be better than what it is replacing. Just eliminating the drunk drivers on the road would halve the accident fatalities. That in itself would be a good thing.
I wanted to illustrate situations that humans deal with all the time but can be difficult for computers. Can we train AI to handle adverse conditions, unexpected obstructs, and untimely situation? Sure we can. But it take time and data, with for unlikely events is scarce.
I worked in ML field. I know the limitations of it. Take Computer Vision, it didn't seem to help Uber's self-driving car that ran over and killed that person. Both radar and lidar are great but have their limits. What if they feed conflicting data to the computer? How does it decide?
I never implied this was on the level of biz logic on a web framework. I have coded some of the most sophisticated ML code and know that it still not ready to drive my car.
Could you code a computer to avoid a child running in the road?
Could you code a computer to stay in it's lane during a blizzard when the lane lines are obscured?
Could you code a computer to steer out of a skid on black ice?
Could you code a computer to drive in pea soup fog when the sensors are blinded?
The difference between human drivers and self-driving cars is hundreds of millions of lines of code that all have to work perfectly.
Hundreds of millions of lines of code? My brain does it like this: f(current situation) -> best action
AI is the same. That's one line of code.
The problem is: 1) deciding what f to use, which is an optimization problem. Easy. 2) what data comprises the "current situation" (cameras, LIDAR, etc.). Hard, and prone to human bias. 3) what do you mean by best (minimize accidents, minimize accidents involving casualties, minimize number of deaths, etc.?). Very Hard.
When you can define 2 and 3 well enough, 1 is easy. That is why AI can be useful.
SFD
One line of code except for the library call.
"f(current situation)" is a couple hundred million lines right there.
"best action" is another couple hundred million lines.
Don't forget the OS. You are not going to run this on Windows. You don't want the blue screen of death right as you are heading for a sharp curve.
The complexity of autonomous vehicles is immense, especially since the general public and regulators are expecting them to be better at making decisions and safer than human drivers. I'd be willing to say that it's orders of magnitude bigger than the difference between reusable and non-reusable rockets.
The difference between reusable and non-reusable rockets is physics and engineering.
The difference between human drivers and self-driving cars is hundreds of millions of lines of code that all have to work perfectly.
Trust me, physics is easier.
Ford> Hey, were not Uber!
I don't know if these "You Tube Moderators" have looked around much, but you can find hateful comments on pretty much any video. Why block comments on just this one?
If YouTube wants to block "hateful comments" why do they allow comments anywhere?
Because the quantity and hatefulness blew the scale?
Are these people new to the Internet?
The ethics board for AI folds because of unethical behavior of the non-artificial intelligent types.
It seems that some at Google think that having an opinion different than their group-think is reason enough to try and silent them.
No tolerance for diversity of thought. No tolerance for different opinions. Hate them because they are different. That is their moral compass?
Is that what we want to teach our AI?
If I answered every email, I would not get any other work done. Most people cc the whole team for every email, creating huge email chains.
If you are answering all your emails then you are not doing any real work, AKA you are a manager.
This is the digital age's equivalent to a paper pusher.
Actually, the second biggest population center in California is San Diego.
Time to pull yesterday's backup tapes. You do have the tapes from yesterday, don't you?
This is just another cases of 'Dumb coders make dumb mistakes'.
News at 11.
I found your argument to be reasonable and well thought out.
You do know that this is Slashdot right?
The reply was meant to be tongue-in-cheek to the argument that the "market economy" and "competition" would push consumer to one company's product vs another's because of the carbon tax.
The new gizmos have to competitive economically or very nearly so for that carbon tax to have that effect. There is promising products and research that do offer lower carbon emissions but not all are competitive costwise with the products they are trying to replace.
That doesn't mean we shouldn't keep trying to reduce our carbon footprint where we can, just that the "green new deal" proposal that all these technologies will magically be ready in a decade is absurd.
If what you say is true then solar power plants would be cheaper than coal power plants, electric cars would be cheaper that gasoline cars, and electric planes would be cheaper to fly than jet aircraft. None of these green technologies have any carbon tax applied to them hence their products will be less expensive. Consumers will flock to them because of the lower prices and carbon emissions will plummet.
WAIT...Reality check...none of that is true.
Apple could order a million of those tiny iPhone screws and it wouldn't take up more than a shelf in space. It could dual source them so it never runs out. One manufacturer in the US and one in China.
This is an easy solvable problem in supply chain management, something that you think Apple would be good at after 40+ years in business. That what happens when you outsource everything. It seems they outsource their business intelligence too. Now their brain-dead management can't figure it out.
Simple, IBM wants cheap skilled employees. Older skilled employees are expensive.
IBM is never going to fix it's skills gap until they value their entire workforce.
Who would want to work there when you are just a cog until you get older or the "hot technology of the day" changes and then they "resource action" you (IBM's term for layoffs).
This is coming from an ex-IBMer
Does your doctor work 100 hour weeks, how about your lawyer, the engineer designing the bridge you will be driving over, or the high rise you will be working in, or the pilot flying the plane you are traveling in? Ask the surgeon if he as performed 99 hours of surgery before he cuts you open for a 'simple one hour surgery. Or the pilot of a plane if he is putting in 100 hours of flying that week?
In all cases, the answer will be no but probably not even close to 100 hrs. But this is expected of software engineers, especially in the gaming industry. Treat them like digital garment workers. Wringing every last ounce out of them. Why should management care about their health, employee burnout, or work-life balance? If you don't like it, they say, "Just quit". They will just find the next naive inexperienced developer to fill the role.
There really is a simple solution to this all. Pay all software engineers by the hour. Overtime paid at time and a half. When the dev costs suddenly triple, management will take notice and stop this insane practice.
Tech geniuses create AI.
Make 500 models, teach it to recognize some 50,000 terms.
AI does HR's job too well.
Executives kill project.
There certainly have been times in history where scientists have disputed with each other in less than sociable ways. Newton and Hooke, Newton and Leibniz, Chandrasekhar and Eddington to name a few. There have even been cases where a more prominent scientist has denied admission to the Royal Society of a scientific rival.
He is wrong, "physics was invented and built by physicists." But he was right, "it's not by invitation". It is not a social club. You don't get a invitation in the mail. You join by achievement, by accomplishment. All this gender talk is a distraction from real physics.
So, if you are in the U.S. and you know Cobol already, you might get a few years of employment out of it. However, such jobs will go overseas, too.
You will have a job for a few years and then they will outsource your position. A few years later you can come back as a consultant to fix the spaghetti code that the overseas 'coders' ran through a blender.
I see it happen all the time.
Master -> Political Correctness Slave -> White Men
Are you saying White Men are now the slaves. Oh the irony!!
Here is some more for you list.
1) Hero / Sidekick
2) Nerd / Wanker
3) Worker / Slacker
4) Villain / Minion
5) Free Thinker / SJW