3 ms·

Asking the Big Questions at Singularity University

3 points by grellas 16y ago

btilly 16y ago
This is something I thought about back in the 90s. Here is a simple thought experiment. Suppose that there was a computer selling for $1,000,000 that was equivalent to a human. With a 5 year amortization schedule, said computer would be equivalent to paying someone $200,000/year. Not really a threat to humans. Fast forward 3 years. After two generations of Moore's Law, said computer now costs $250,000, and can profitably replace people worth $50,000/year. Previous generations of technology always had the result that there are other things humans can be trained to do more easily than you can build machines to do it, so humans move. But there is nothing you can train displaced people to do that the computer can't do as well. What happens? Fast forward another 3 years. Most people are unemployed because they can't do anything that those with money can't get done with machines. In theory there is enough excess capacity for everyone. In practice investors will demand returns and efficiency, and attempts at economic redistribution have never succeeded. There are a lot of arguments against this. A common one is based on the concept of competitive advantage and the benefits of free trade. Adding robots is therefore good. However that argument has a major flaw. It is true that with 2 parties, if you open up free trade then both benefit. It is true that with 3 parties, opening up free trade leaves everyone better off. However it is also possible (and fairly common) that free trade with 3 parties works to the advantage of two, and the disadvantage of the third. (For a real world example, look at the long-term economic decline along the silk road after Europe began trading directly with Asia via ships.) A second one is based on history. It is true that every advance in technology has left people better off. But that is because there was always something that people had a comparative advantage over machines at that people really wanted. And the cause of our comparative advantage is our intelligence. If machines can supply that, then we cannot demand a salary that is greater than the value of the resources needed to do the same job with machines. Once that happens, Moore's Law is _not_ our friend any more.
- ugh 16y ago
  It seems obvious to me that if we can be a lot more productive with a lot less human work we will have to change the way our economy works. That’s a challenge and we might very well fail but I don’t think it’s impossible. Should we succeed the elimination of the need for humans to work would obviously be very positive [+]. You seem to me a bit to quick at brushing that possibility aside. I think that it is already not true that attempts at economic redistribution have never succeeded. I know that Germany and many other European nations already redistribute money on a relatively massive scale (unemployment benefits, child benefits, and so on). My knowledge about the US is somewhat limited but I would bet that there are at least some redistributive policies in place. [+] Barring other possible problems with AI like it killing all humans.

3 points by grellas 16y ago

This is something I thought about back in the 90s. Here is a simple thought experiment. Suppose that there was a computer selling for $1,000,000 that was equivalent to a human. With a 5 year amortization schedule, said computer would be equivalent to paying someone $200,000/year. Not really a threat to humans. Fast forward 3 years. After two generations of Moore's Law, said computer now costs $250,000, and can profitably replace people worth $50,000/year. Previous generations of technology always had the result that there are other things humans can be trained to do more easily than you can build machines to do it, so humans move. But there is nothing you can train displaced people to do that the computer can't do as well. What happens? Fast forward another 3 years. Most people are unemployed because they can't do anything that those with money can't get done with machines. In theory there is enough excess capacity for everyone. In practice investors will demand returns and efficiency, and attempts at economic redistribution have never succeeded. There are a lot of arguments against this. A common one is based on the concept of competitive advantage and the benefits of free trade. Adding robots is therefore good. However that argument has a major flaw. It is true that with 2 parties, if you open up free trade then both benefit. It is true that with 3 parties, opening up free trade leaves everyone better off. However it is also possible (and fairly common) that free trade with 3 parties works to the advantage of two, and the disadvantage of the third. (For a real world example, look at the long-term economic decline along the silk road after Europe began trading directly with Asia via ships.) A second one is based on history. It is true that every advance in technology has left people better off. But that is because there was always something that people had a comparative advantage over machines at that people really wanted. And the cause of our comparative advantage is our intelligence. If machines can supply that, then we cannot demand a salary that is greater than the value of the resources needed to do the same job with machines. Once that happens, Moore's Law is _not_ our friend any more.

ugh 16y ago
It seems obvious to me that if we can be a lot more productive with a lot less human work we will have to change the way our economy works. That’s a challenge and we might very well fail but I don’t think it’s impossible. Should we succeed the elimination of the need for humans to work would obviously be very positive [+]. You seem to me a bit to quick at brushing that possibility aside. I think that it is already not true that attempts at economic redistribution have never succeeded. I know that Germany and many other European nations already redistribute money on a relatively massive scale (unemployment benefits, child benefits, and so on). My knowledge about the US is somewhat limited but I would bet that there are at least some redistributive policies in place. [+] Barring other possible problems with AI like it killing all humans.