For the sake of this comment, I’m assuming a world (like what I imagine this post to be describing) where AI development is smooth but quick, and AI ends up aligned enough that humans continue to survive and retain effective control of the future through being the users of AI systems, and we continue in a regime where things like money and property rights and laws made by human governments are meaningful. I don’t think that’s the most likely path, but I like this discussion of what it would entail.
The phase transition you’re describing, physical capital no longer being productivity-limited by human capital, makes a lot of sense in that world. What that does to GDP, or to particular prices, is a lot less clear to me. Mainly because the price of something is, ultimately, set by markets based on willingness (of those with money) to pay. At first, someone needs to build extra physical capital, and in that phase, production is still rate limited, and prices get bid up. But, this phase probably doesn’t last long enough to affect the human capital stock or cost of labor much, and mostly looks like a windfall for manufacturers, shortly (IDK how long this phase actually lasts, I would guess no more than a couple of years) before they become obsolete. By which I mean, capital becomes so abundant and self replicating that prices collapse, and the companies either go out of business or their revenue becomes miniscule (but if they have no employees left they might technically stay in business).
Along the way, prices of every other good or service also collapse due to automation, leaving (almost) everyone unemployed and unemployable, but possibly leaving companies intact and theoretically profitable?
I would anticipate a major divergence of nominal GDP growth from real GDP growth, where nominal prices collapse despite real wealth rapidly rising. This plausibly causes almost anyone with debt to default, except governments borrowing in their own currency. At which point their assets go to creditors, who may find there’s no one able to buy (or interested in buying) them at more than a miniscule fraction of their previous nominal price. Repeat as needed until we know who is actually solvent and who owns all the things.
What happens next? The good outcomes may be things like “Society actually realizes this is a big deal, and that all this tech is useless unless we make it enrich and empower the vast majority of people, and quit worrying about things like benefitting Those People Who Don’t Deserve It for whatever reason, or about not benefitting everyone exactly equally.” The bad outcomes range from “Let most people starve amidst plenty” to “Incredibly destructive wars and revolts over resources” to “The remaining wealthy handful near-instantly amass unassailable armies of autonomous weapons and use them to retain control.”
For the sake of this comment, I’m assuming a world (like what I imagine this post to be describing) where AI development is smooth but quick, and AI ends up aligned enough that humans continue to survive and retain effective control of the future through being the users of AI systems, and we continue in a regime where things like money and property rights and laws made by human governments are meaningful. I don’t think that’s the most likely path, but I like this discussion of what it would entail.
The phase transition you’re describing, physical capital no longer being productivity-limited by human capital, makes a lot of sense in that world. What that does to GDP, or to particular prices, is a lot less clear to me. Mainly because the price of something is, ultimately, set by markets based on willingness (of those with money) to pay. At first, someone needs to build extra physical capital, and in that phase, production is still rate limited, and prices get bid up. But, this phase probably doesn’t last long enough to affect the human capital stock or cost of labor much, and mostly looks like a windfall for manufacturers, shortly (IDK how long this phase actually lasts, I would guess no more than a couple of years) before they become obsolete. By which I mean, capital becomes so abundant and self replicating that prices collapse, and the companies either go out of business or their revenue becomes miniscule (but if they have no employees left they might technically stay in business).
Along the way, prices of every other good or service also collapse due to automation, leaving (almost) everyone unemployed and unemployable, but possibly leaving companies intact and theoretically profitable?
I would anticipate a major divergence of nominal GDP growth from real GDP growth, where nominal prices collapse despite real wealth rapidly rising. This plausibly causes almost anyone with debt to default, except governments borrowing in their own currency. At which point their assets go to creditors, who may find there’s no one able to buy (or interested in buying) them at more than a miniscule fraction of their previous nominal price. Repeat as needed until we know who is actually solvent and who owns all the things.
What happens next? The good outcomes may be things like “Society actually realizes this is a big deal, and that all this tech is useless unless we make it enrich and empower the vast majority of people, and quit worrying about things like benefitting Those People Who Don’t Deserve It for whatever reason, or about not benefitting everyone exactly equally.” The bad outcomes range from “Let most people starve amidst plenty” to “Incredibly destructive wars and revolts over resources” to “The remaining wealthy handful near-instantly amass unassailable armies of autonomous weapons and use them to retain control.”