Spot on on not needing to work IIF we get redistribution right - both within and across countries (assuming these still exist/assuming a non-everything-altering takeoff).
Given how rich nations treat poor nations today, I'm unconvinced about us getting redistribution globally right even in a world of more abundance than today, though who knows.
Saying that goods will be cheap doesn't say cheap relative to what. The relevant exchange rate is between human labor and the basket of goods required to sustain a human (food, housing, etc). AI will devalue both, but it seems to me that it'll devalue labor faster. You can see that by looking at energy equivalent: it's very possible that using AI to do an equivalent of your day's intellectual work will require just a tiny amount of electricity to generate some tokens, but growing enough food to feed you for a day will require more electricity than that, because growing real physical apples can only be optimized so far. If that's so, then people might well end up unable to earn a living, even in the midst of AI abundance.
As a fallback you mention investments and redistribution. But not everyone has investments (lots of people have negative net worth), and redistribution to a group that has no leverage is unlikely to be stable in the long run. At some point the handout might get taken away and we'll have no recourse. See social programs being cut now.
There's a class that lives this way today.
Those on the right side of credit, who borrow cheaply and collect interest from everyone else, can asset-strip the other side at will.
I don't see why AGI changes that. If robots take over what the aristocracy actually depends on (extraction, manufacturing, medical, security), what remains are the servile roles they'd prefer a human to fill. The OP assumes the provider's alternative is leisure. For a debtor the alternative is default and an inability to access basic needs (shelter, food, clothing, transportation, social connection, etc). Credit availability and interest rates then become the rationing mechanism.
So provided that AI tools are aligned to the interests of the aristocracy and can replace labor, the status quo persists or intensifies.
For as long as superintelligence doesn't break the current paradigm, maintaining enough prosaic alignment/control remains plausible, which is also a requirement for a normal-ish world. This stops being the case either suddenly (after some breakthrough), or after the industrial explosion gains enough momentum to brute-force superintelligence with the sheer amount of compute.
The way things are going, humans are very likely to lose control of almost all of the resources (or more). If the AIs permit non-extinction, even the remaining crumbs of resources will still make work instrumentally meaningless, but that's an importantly different situation from owning the future (which humanity collectively still does, but this fact is slipping away).
The kind of industrial explosion that creates overwhelming wealth also creates an overwhelming advantage for the AIs, possibly with a "peaceful" gradual disempowerment AI victory that takes away the future without too much overt conflict, at some point after the AIs are handed control over most of the infrastructure, and this infrastructure grows to many times the legacy human world.
FWIW, I explain in How Econ 101 makes us blinder on trade, morals, jobs with AI – and on marginal costs what fallacy according to my experience leads fellow economists to dismiss salary shortage scenarios.
I think there is something fundamentally wrong with the notion that once we have powerful AI coupled with robots that everything is going to be available at prices too cheap to meter. While it may be possible to assemble goods without workers, at some point you have to generate the energy to make all of this possible. There are a number of problems with that:
Using more energy will, by itself, cause the planet to heat up. Collecting "free" solar power to strip-mine the rare earth deposits needed to create more solar panels is an inefficient process, generating waste heat at every step along the way. We already know that densely populated areas are heat islands that are measurably warmer than their general climate would suggest.
There is also no known way to implement the heavy industry that underpins all of this in a way that doesn't lead to large amounts of pollution. Mining sends all sorts of unpleasant materials into the surrounding air, soil, and water. Running a blast furnace not only produces toxic slag but also generates a fair amount of carbon monoxide due to the chemistry required. Rinse and repeat for most other large-scale manipulation of atoms.
Then there are the things that cannot be made that much better. Unless the AI develops some miracle way to grow fields of wheat in a week without huge increases in the water and fertilizer required, there are still going to be limits on food. The higher up the food chain you go the worse this gets. We are already raping the seas for all commercially viable sea creatures; coming up with faster ways to harvest the remaining tuna doesn't make any more of them. (And farming fish produces a lower-quality product with lots of associated pollution.)
Finally, while it may be that AI is truly transformative and overcomes all of the above objections, it's always a good idea to look at history for a clue as to how humans might manage this societal upheaval. Looking at the largest-scale changes in human history doesn't provide happy reading:
It is true that all but the very poorest people today live better than the very richest hunter-gatherers (medical care alone makes a huge difference in life expectancy and enjoyment), so humanity as a whole has gained from all these advancements. However, through all of this surplus we have always had a small number of ruthless people who rose to the top and grabbed everything they could. I find it nearly impossible to believe that Elon Musk and company are going to suddenly distribute their power and access to benefits to the rest of the world just because AI worked out really well.
Many economists take a relaxed view of employment post-AGI. Even once AI systems with suitable robotic actuators can perform most white-collar work and manual labor, something will remain scarce, and comparative advantage guarantees that human labor is worth hiring for something at some wage. So people will still have jobs, spending most of their waking hours at work, and the economy will look broadly like today's, including property rights, wage labor, and political institutions protecting these.
I don't think this holds up. It is not obvious that current economic or political structures survive the arrival of AGI at all. Powerful AI systems (or even simply AI-enabled human dictatorships) need no more respect our rights than Stalin respected those of the kulaks. But set those concerns aside and grant an optimistic future where power—and access to the fruits of automation—remain broadly distributed. Even then, mass wage employment does not follow, though I do expect former workers to be fine.
The basic issue is this: why would anyone choose to be employed? People today sell their labor because they need to buy food, shelter, clothes, and other material goods. Post-AGI those goods become abundant and cheap, because they can be produced by robots and AI systems that reproduce themselves freely and rapidly. Anyone with some investments owns a share of physical capital that reproduces itself, and their income grows every year whether or not they lift a finger. Even without investments, physical capital would soon be so abundant that little work would be needed to acquire it, and a modicum of redistribution could provide wealth to everyone.
Comparative advantage does not overcome this. It guarantees that humans can still add some value even once autonomous systems have an absolute advantage at everything, but it does not mean that value is greater than that of their leisure. Human workers are not like the horses that disappeared once their wages no longer covered their subsistence. They are like retirees who have enough invested to live comfortably without working.
The motivation to work must therefore come from something that remains scarce despite this material abundance. But work consumes our time—perhaps our most precious resource—so what other scarce resources could justify anything approaching full-time employment?
Other goods that human labor cannot provide may still remain scarce. There will always be a limited number of tropical islands and Gutenberg Bibles, and there can be only one world's-largest banana statue. But working does not make more of them. It only changes who holds them, and the person holding one has nothing to gain from your labor unless it is a service they want. So these goods alone cannot provide the basis of a wage economy.
This leaves the services that truly require a human to provide. Purported examples include babysitters, ballerinas, sex workers, and waiters. It is debatable in each case whether, post-AGI, people would in fact still demand that a human do the job rather than using AI substitutes that are cheaper and plausibly superior along many axes. But even taking for granted an intrinsic demand for humans in these roles, I don't think these jobs can support anything resembling the current labor market.
The problem is the gap between what buyers would pay and what providers would accept. People would consume plenty of these services if they were very cheap, and would supply them if paid a great deal. But the real cost of an hour of human service is an hour of someone's leisure. The buyer must therefore value the service being performed by a human—rather than by an abundant AI substitute—enough to compensate the provider for giving up that leisure. Would you wait tables for someone so that they would wait tables for you, or babysit someone else's child so that they could babysit yours?
The appetite for such trades is not zero. Differences in talent and taste can create some gains from trade: a talented cook might trade dinners for babysitting, and a few performers will always draw crowds. But once machines supply the skill, what a human adds is their humanity, and people differ far less in that than in skill. So the trades are mostly time for time or prestige for prestige, and the hour received is worth little more than the hour given. That thin margin has to cover the work itself and the status cost of serving someone else. And where the provider's humanity matters most, it is usually because the service is intimate and social—the very qualities that make it awkward to organize through impersonal wage exchange. This looks a lot more like trading favors among friends than the daily grind.
Indeed, we have many examples of people whose material needs are met and who have time on their hands: retirees, college students on stipends, the inherited rich, and aristocrats of all stripes. They spend their time competing for status and rank, keeping up hobbies and prestige projects, hanging out with friends and doing each other favors. Many of them work hard. Some even hold jobs, but while their material needs are met they generally neither depend on wage labor nor regard employment as a prerequisite for a fulfilling life.
Wealthy people today often continue working, because they value work, and the status it brings, for its own sake. That is unsurprising in a society where work is the predominant form of status, and even so the really rich are mostly not doing anything that resembles normal wage labor. We could keep work at the center of status post-AGI, the way some Gulf states distribute oil rents through public-sector jobs that exist mainly to pay a salary. But history shows that other options are available, and it is not obvious why we would bother with fake jobs to continue a 20th century charade. We should not imagine post-AGI humans as workers whose jobs were automated away, but as aristocrats who never needed them.