LogFAQs > #952841391

LurkerFAQs, Active DB, DB1, DB2, DB3, DB4, DB5, DB6, DB7, Database 8 ( 02.18.2021-09-28-2021 ), DB9, DB10, DB11, DB12, Clear

Topic List	Page List: 1
Topic	Controversial Opinion #4: Automation
darkknight109 04/14/21 9:07:13 PM #134:	@LinkPizza LinkPizza posted... Plus, if there are certain things people do, many may want to be compensated in some forms. If people want compensation, they'll need to do it better than an AI. In a fully automated future, there's nothing they can do that will be better than an AI. LinkPizza posted... And money can (and will) exist because we will still need currency or some sort for many things. For example, buying and selling land. And animals for things like farming, since they are 3D printed. Plus, if there are certain things people do, many may want to be compensated in some forms. And even if you could make anything you wanted, you would need the materials to do so. And those materials would most likely cost money (which this already happens)Not to mention things that may require finite resources. They cant just give all those out for free And restricting those means that we wouldnt be able to just make what we want when we want Congratulations, you've caught up to where I was when I started this side tangent. You've struck upon the fact that there's a fundamental issue, which is that there is still scarcity of some things (land and limited raw materials), which necessitates some form of exchange. However, you still haven't realized that without the ability for humans to operate a job market (why would anyone hire you for work when they can get a robot that will do the job better and without demanding a wage?), there is no way to earn money and, therefore, no way for money to exist as a tool of exchange. This is one of those fundamental problems I've pointed out that we need to solve. LinkPizza posted... Yeah. But wed have to program them to do that first. Because even if they can learn, theyd have to learn from something else. And wed probably start making them before the AI could learn how to make them. This is actually quite incorrect. AI, even today, program themselves better than we can. We don't actually know how the most advanced AI's programming works, because it was the result of a self-programming process that produced an incredibly complex result that no programmer in the world can replicate. As a very simplistic example, let's say you were training an image-recognition program to identify whether a picture contains a cat or not. Trying to program what a "cat" is into an AI is an exercise in futility, because it is almost impossible to do manually. Instead, you get an AI to do it instead (which seems counter-intuitive, since if you can't tell a program what a cat is, how can you tell an AI to make you a program that can tell what a cat is, but it actually does work). The AI takes an image recognition program and basically continually modifies its parameters while testing it against millions of calibration images where it is already known if they contain cats or not. Once it fine-tunes the program to the point where it can correctly identify the cats in each calibration image, it then collects new data to use as test data (CAPTCHAs are frequently used for this purpose) to see if the program works on non-control data. It iterates this process over and over until it can flawlessly distinguish what is and isn't a cat. Hence, the AI succeeds in making a program that is difficult-to-impossible for a human to make. The same is true of, say, chess-playing robots (how do you program an AI to outsmart a grandmaster when you yourself are not a grandmaster?) or music-creating robots (how do you program an AI to create music if you aren't musically inclined yourself?) or, really, almost any AI. LinkPizza posted... It's something that would need to be perfect every time, though. While also being somewhat fast. They don't have to be perfect every time, though, because humans aren't perfect every time. Heck, they technically don't even have to be as good as us or as fast as us. A robot that does things half as quickly and half as accurately as a human is still cost effective if it's 1/10th the cost. To management that only cares about dollars and cents, they'll happily make that trade because it puts more money in their pockets. LinkPizza posted... Especially considering that I believe the self-driving ones go slower than normal buses (based on the videos I saw), and will probably have a little trouble keeping up with the routes. Self-driving vehicles can go as fast as you want them to. Their reaction times are far, far faster than even the best humans. You'll hit the limits of the vehicle before you hit the limits of the AI driving it. Self-driving cars tend to go slower than humans on the road because they obey the speed limit, where most humans don't, and drive in a very accommodating, non-aggressive manner, where many humans don't. LinkPizza posted... As for cost efficiency, that means we wont see them for a while, meaning most buses wont be fully autonomous (unless they dont care about the handicapped population). I never said this was happening tomorrow. But the writing is on the wall and those fully automated vehicles are going to be taking over faster than you think. LinkPizza posted... I still see a never ending line or automation, though. Thats where the problem is. How? I literally pointed out how you do this with a finite number of automatons - as many or as few as you like. --- Kill 1 man: You are a murderer. Kill 10 men: You are a monster. Kill 100 men: You are a hero. Kill 10,000 men, you are a conqueror! <cite>darkknight109 posted [p:952841391] in Controvers... [t:79384298]...</cite> <quote>@LinkPizza LinkPizza posted... </cite> <quote>Plus, if there are certain things people do, many may want to be compensated in some forms.</quote>If people want compensation, they'll need to do it better than an AI. In a fully automated future, there's nothing they can do that will be better than an AI. LinkPizza posted... </cite> <quote>And money can (and will) exist because we will still need currency or some sort for many things. For example, buying and selling land. And animals for things like farming, since they are 3D printed. Plus, if there are certain things people do, many may want to be compensated in some forms. And even if you could make anything you wanted, you would need the materials to do so. And those materials would most likely cost money (which this already happens)Not to mention things that may require finite resources. They cant just give all those out for free And restricting those means that we wouldnt be able to just make what we want when we want</quote>Congratulations, you've caught up to where I was when I started this side tangent. You've struck upon the fact that there's a fundamental issue, which is that there is still scarcity of some things (land and limited raw materials), which necessitates some form of exchange. However, you still haven't realized that without the ability for humans to operate a job market (why would anyone hire you for work when they can get a robot that will do the job better and without demanding a wage?), there is no way to earn money and, therefore, no way for money to exist as a tool of exchange. This is one of those fundamental problems I've pointed out that we need to solve. LinkPizza posted... </cite> <quote>Yeah. But wed have to program them to do that first. Because even if they can learn, theyd have to learn from something else. And wed probably start making them before the AI could learn how to make them.</quote>This is actually quite incorrect. AI, even today, program themselves better than we can. We don't actually know how the most advanced AI's programming works, because it was the result of a self-programming process that produced an incredibly complex result that no programmer in the world can replicate. As a very simplistic example, let's say you were training an image-recognition program to identify whether a picture contains a cat or not. Trying to program what a "cat" is into an AI is an exercise in futility, because it is almost impossible to do manually. Instead, you get an AI to do it instead (which seems counter-intuitive, since if you can't tell a program what a cat is, how can you tell an AI to make you a program that can tell what a cat is, but it actually does work). The AI takes an image recognition program and basically continually modifies its parameters while testing it against millions of calibration images where it is already known if they contain cats or not. Once it fine-tunes the program to the point where it can correctly identify the cats in each calibration image, it then collects new data to use as test data (CAPTCHAs are frequently used for this purpose) to see if the program works on non-control data. It iterates this process over and over until it can flawlessly distinguish what is and isn't a cat. Hence, the AI succeeds in making a program that is difficult-to-impossible for a human to make. The same is true of, say, chess-playing robots (how do you program an AI to outsmart a grandmaster when you yourself are not a grandmaster?) or music-creating robots (how do you program an AI to create music if you aren't musically inclined yourself?) or, really, almost any AI. LinkPizza posted... </cite> <quote>It's something that would need to be perfect every time, though. While also being somewhat fast.</quote>They don't have to be perfect every time, though, because humans aren't perfect every time. Heck, they technically don't even have to be as good as us or as fast as us. A robot that does things half as quickly and half as accurately as a human is still cost effective if it's 1/10th the cost. To management that only cares about dollars and cents, they'll happily make that trade because it puts more money in their pockets. LinkPizza posted... </cite> <quote>Especially considering that I believe the self-driving ones go slower than normal buses (based on the videos I saw), and will probably have a little trouble keeping up with the routes.</quote>Self-driving vehicles can go as fast as you want them to. Their reaction times are far, far faster than even the best humans. You'll hit the limits of the vehicle before you hit the limits of the AI driving it. Self-driving cars tend to go slower than humans on the road because they obey the speed limit, where most humans don't, and drive in a very accommodating, non-aggressive manner, where many humans don't. LinkPizza posted... </cite> <quote>As for cost efficiency, that means we wont see them for a while, meaning most buses wont be fully autonomous (unless they dont care about the handicapped population).</quote>I never said this was happening tomorrow. But the writing is on the wall and those fully automated vehicles are going to be taking over faster than you think. LinkPizza posted... </cite> <quote>I still see a never ending line or automation, though. Thats where the problem is.</quote>How? I literally pointed out how you do this with a finite number of automatons - as many or as few as you like. --- Kill 1 man: You are a murderer. Kill 10 men: You are a monster. Kill 100 men: You are a hero. Kill 10,000 men, you are a conqueror!</quote> ... Copied to Clipboard!
Topic List	Page List: 1