Why on Earth would human cost of living become cheaper in terms of human working hours? Humans are very inefficient users of resources compared to future AI.
Here’s an analogy: if we introduced the same number of horses to London today as there were in 1900, could they all afford luxurious horse-lives by selling horse-labor? Nope! Most of them wouldn’t even be able to afford rent. London’s economy has way more efficient things to do than keeping lots of horses for their labor.
I wrote a post explaining this argument in more detail. The conclusion is that we must get AI values right, market dynamics won’t save us.
Hmmm, I didn’t find your linked post particularly related. Maybe I’m just not smart enough to understand the connection. It seems like you are talking about post-singularity ai, and I am certainly talking about pre-singularity. Sure, if the ai is actually looking at the atoms in our bodies as resources to fuel it’s paperclip factory and has the power to take them, there’s no cost of living adjustment that’s going to offset being turned into paperclips. But the whole concept of the singularity is that it’s kind of pointless to speculate about what happens afterwards anyway.
Given that I’m talking about near-term ai, all I’m trying to say is ai will take human jobs if and only if they reduce the input cost of production. That’s a big downside for the person laid off, but a small upside multiplied many times over for everyone else. As the cost of goods falls, ceteris paribus people’s real wages rise (tautologically). As ai become more efficient and capable of doing more tasks, prices tend towards pure resource cost for production of any given good. Ceteris paribus, under such conditions real wages would be enormously higher than nominal wages today.
This seems obvious to me, but I don’t hear much about this upside from anyone in the ai is taking our jobs conversation.
My assertion, though you’re right to point out it’s just conjecture, is that even when nominal wages fall due to reduced working hours, this will be offset by huge increases in buying power.
I find your horse analogy flawed in 2 respects:
in my scenario costs of inputs are reduced, whereas in London the cost of sufficient land to keep a horse and stable are vastly increased. If it was almost free to keep horses in London, I surmise they wouldn’t have to do very much to earn their keep, and plenty would.
in fact, we’ve see a huge increase in previously working and livestock species being kept as pets in both rural and urban environments as the cost of feeding and housing them fell, and all they offer in return is a vauge sense of companionship.
Why on Earth would human cost of living become cheaper in terms of human working hours? Humans are very inefficient users of resources compared to future AI.
Here’s an analogy: if we introduced the same number of horses to London today as there were in 1900, could they all afford luxurious horse-lives by selling horse-labor? Nope! Most of them wouldn’t even be able to afford rent. London’s economy has way more efficient things to do than keeping lots of horses for their labor.
I wrote a post explaining this argument in more detail. The conclusion is that we must get AI values right, market dynamics won’t save us.
Hmmm, I didn’t find your linked post particularly related. Maybe I’m just not smart enough to understand the connection. It seems like you are talking about post-singularity ai, and I am certainly talking about pre-singularity. Sure, if the ai is actually looking at the atoms in our bodies as resources to fuel it’s paperclip factory and has the power to take them, there’s no cost of living adjustment that’s going to offset being turned into paperclips. But the whole concept of the singularity is that it’s kind of pointless to speculate about what happens afterwards anyway.
Given that I’m talking about near-term ai, all I’m trying to say is ai will take human jobs if and only if they reduce the input cost of production. That’s a big downside for the person laid off, but a small upside multiplied many times over for everyone else. As the cost of goods falls, ceteris paribus people’s real wages rise (tautologically). As ai become more efficient and capable of doing more tasks, prices tend towards pure resource cost for production of any given good. Ceteris paribus, under such conditions real wages would be enormously higher than nominal wages today.
This seems obvious to me, but I don’t hear much about this upside from anyone in the ai is taking our jobs conversation.
My assertion, though you’re right to point out it’s just conjecture, is that even when nominal wages fall due to reduced working hours, this will be offset by huge increases in buying power.
I find your horse analogy flawed in 2 respects:
in my scenario costs of inputs are reduced, whereas in London the cost of sufficient land to keep a horse and stable are vastly increased. If it was almost free to keep horses in London, I surmise they wouldn’t have to do very much to earn their keep, and plenty would.
in fact, we’ve see a huge increase in previously working and livestock species being kept as pets in both rural and urban environments as the cost of feeding and housing them fell, and all they offer in return is a vauge sense of companionship.