3 ms·

The video is a well articulated argument, but there are two large gaps in reasoning that are common with this line of thought. Grey ignores the idea that techn

The video is a well articulated argument, but there are two large gaps in reasoning that are common with this line of thought. Grey ignores the idea that technology also allows for normal people to do jobs that used to require highly skilled specialists. A big part of this is so called 'brain' labor. A banker before computers had to have amortization tables manually calculated. Now the personal banker who does your HELOC probably doesn't even know basic algebra. If we ran technology in reverse and took away the calculators we would end up with fewer banking jobs not more. Because every banker would have to be highly specialized and costly. The "factory jobs" were originally an invention of technology. A method to take a large number of workers + automation to beat artisans and specialists. If Watson is doing diagnosis, that means that more people with less training can be effective 'doctors'. Which brings me to my second point. Often when things become more efficient we end up using more of them overall. Jevon's paradox applied to labor. Think of what it cost to transport 20 tons of cargo from NY to SF before trucking and standardized containers. How many man hours were involved? How many different specialized jobs? Now done by one guy with a truck replacing them all. Yet the number of people employed overall in transportation went up not down. IMHO the best example is IT. Think about the server admin ratios in the early days. Perhaps a team of 10 to maintain a single mainframe. Now we have 10 guys in a data center managing 50,000 servers. Yet the overall number of people employed in IT went up not down. As processes become more efficient and cheap we consume more of it at an increasing rate.

CaptSpify 10y ago
> Grey ignores the idea that technology also allows for normal people to do jobs that used to require highly skilled specialists. Computers are much better at highly specialized jobs than people are. And the amount of things they are not good at is slimming down all the time. > If Watson is doing diagnosis, that means that more people with less training can be effective 'doctors'. And it'll eventually improve to the point that we won't need 'doctors'. It'll get to the point that your average person can be their own 'doctor', because most of the work is shelled out to a computer. > Yet the number of people employed overall in transportation went up not down. Because even with more automated, cheaper shipping, we still had jobs that robots couldn't do. But more of those jobs are disappearing every day. I think your logic is skipping over the fact that computers don't have hard limits like humans do. If a computer can't do something, just give it a handful of engineers, and some time, and it'll eventually be able to do that. Once a computer gets better at a task than humans, there's no reason to ever go back to humans. It's a one way street. Sure, computers can't do everything we do now, but they will. It's only a matter of time.
- johngalt 10y ago
  You're conflating technology that is available today with challenges which occur only after we have AGI. When those are different problems with different time horizons. Post AGI. Obviously robots do everything. Post AGI it is the technology singularity as we have recursively improving superintelligences. Unemployment in this scenario is a trivial issue. Strong friendly AI would solve so many problems that unemployment would be one of those historical footnotes. Concerning ourselves with 'what happens when computers are better than humans at everything?' is akin to asking 'what should we do after the rapture?' Prior to AGI, human + computer is the norm for most tasks. Similar to my banker example. You'll have specialized computers augmenting human productivity. In each case driving costs down and in most cases usage up. Specialized computers will make it easier for humans to do jobs they couldn't do before. Even if truck driving becomes fully automated there will still be humans in the loop somewhere. It will just mean a more efficient ratio of man hours per ton shipped. Which is not a fundamentally different issue than dealing with any other efficiency increase. > your logic is skipping over the fact that computers don't have hard limits like humans do. If a computer can't do something, just give it a handful of engineers, and some time, and it'll eventually be able to do that. Once a computer gets better at a task than humans, there's no reason to ever go back to humans. It's a one way street. I don't deny that. Merely that we've been headed down this one way street for a while and the houses keep getting nicer. I keep hearing that Efficiency Avenue will take us towards human suffering, but we've been heading this direction for centuries and never arrived there. In fact everything about what we have seen so far says we should travel as fast as we can down that one way street.

CaptSpify 10y ago

> Grey ignores the idea that technology also allows for normal people to do jobs that used to require highly skilled specialists. Computers are much better at highly specialized jobs than people are. And the amount of things they are not good at is slimming down all the time. > If Watson is doing diagnosis, that means that more people with less training can be effective 'doctors'. And it'll eventually improve to the point that we won't need 'doctors'. It'll get to the point that your average person can be their own 'doctor', because most of the work is shelled out to a computer. > Yet the number of people employed overall in transportation went up not down. Because even with more automated, cheaper shipping, we still had jobs that robots couldn't do. But more of those jobs are disappearing every day. I think your logic is skipping over the fact that computers don't have hard limits like humans do. If a computer can't do something, just give it a handful of engineers, and some time, and it'll eventually be able to do that. Once a computer gets better at a task than humans, there's no reason to ever go back to humans. It's a one way street. Sure, computers can't do everything we do now, but they will. It's only a matter of time.

johngalt 10y ago
You're conflating technology that is available today with challenges which occur only after we have AGI. When those are different problems with different time horizons. Post AGI. Obviously robots do everything. Post AGI it is the technology singularity as we have recursively improving superintelligences. Unemployment in this scenario is a trivial issue. Strong friendly AI would solve so many problems that unemployment would be one of those historical footnotes. Concerning ourselves with 'what happens when computers are better than humans at everything?' is akin to asking 'what should we do after the rapture?' Prior to AGI, human + computer is the norm for most tasks. Similar to my banker example. You'll have specialized computers augmenting human productivity. In each case driving costs down and in most cases usage up. Specialized computers will make it easier for humans to do jobs they couldn't do before. Even if truck driving becomes fully automated there will still be humans in the loop somewhere. It will just mean a more efficient ratio of man hours per ton shipped. Which is not a fundamentally different issue than dealing with any other efficiency increase. > your logic is skipping over the fact that computers don't have hard limits like humans do. If a computer can't do something, just give it a handful of engineers, and some time, and it'll eventually be able to do that. Once a computer gets better at a task than humans, there's no reason to ever go back to humans. It's a one way street. I don't deny that. Merely that we've been headed down this one way street for a while and the houses keep getting nicer. I keep hearing that Efficiency Avenue will take us towards human suffering, but we've been heading this direction for centuries and never arrived there. In fact everything about what we have seen so far says we should travel as fast as we can down that one way street.