I wonder about that too, if we have robots who do jobs how will people be able to pay for stuff?
This happens now with outsourcing - effectively, jobs leave the market but companies still exist and their customers still exist, but some customers no longer have income to pay for goods and services.
I think you'll see robots doing the dangerous, heavy, undesirable jobs. Sorting recyclables out of garbage, mining coal, underwater welding, arctic trucking, crab fishing. The question is, will it ever get to the point where a robot is cheaper and more efficient than a human at other tasks, like sales? Well there are vending machines.
Similarly, a librarian has a lot of useful knowledge but computers are getting better and better at doing their job. A computer teacher could spot when a pupil isn't grasping a concept, offer some options that describe what the pupil may be feeling from which he can choose, and change its lesson to conform to those problems. A program should be able to identify, for example, if a math student doesn't grasp a certain concept and needs to go back and do some remedial learning. A computer program linked to a 3D projection and sensors should be able to teach you a martial art. In these cases you're dealing with a lot of small human cues, language complexities, and so the robot needs to be very well-developed. But once it's developed making copies is cheap, so it puts a lot of teachers out of a job.
A self-driving car should be able to replace a cabbie. The car would be more expensive, but once the costs of developing the technology are paid off actually producing them is not so expensive. Also the car can hold four passengers, meaning it can network with the service and pick up more fares along the way, avoid heavy traffic and construction detours, and interface fully with the automated driving road system. At some point having a driver in the cab is going to stop making sense.
But let's say you have a technician whose job it is to repair and replace the robot teacher or the self-driving car. It's reasonable to assume that the house your teacher is in is too strange and complex for a robot to navigate. And what company will send a robot which can be stolen and tampered with to a work site? A car may be more standardized, so that once it's on the lift the robot mechanic can do everything. But, barring a mythic singularity, you still need humans to actually create stuff.
Some human is going to be an engineer, using complex computer tools, to design a new car. The robots on the assembly line need changes to their tools and they need to be reprogrammed to the new manufacturing process. You still need people involved.
Where does the material come from for these robots and car parts? You need prospecting, which could be aided by robots, but isn't really a thing you can do from orbit AFAIK. The metal must be mined and processed, and that mining equipment needs technicians to keep it working. Sure you could have a robot that fixes the mining equipment, and it's able to fix other repair-bots, but there are always mistakes and unexpected hazards for which you need a human around just in case. You don't want the savings of a few peoples' wages to be what causes your whole operation to shut down for weeks at a time?
//
Establishing this infrastructure is a problem. Right now we have a lot of people who can do work, and cheaply, while developing these robots is expensive and installing them doesn't make economic sense.
But let's say you've established such an infrastructure that robots do most of the work. Since sci-fi writers typically are either way too conservative or way too liberal with their predictions, one might predict that we would have a computer network spanning the globe by 2050 and another might predict a moon-base by 2001. I'd say this level of robotic infrastructure wouldn't occur even in a small rich country before 2100, and then only because (A) there is a dramatic drop in population, or (B) robotics becomes so cheap it's worthwhile compared to human labor.
Regardless, you will have people who own this infrastructure, or it will be owned in common, perhaps by the government.
If it's government-controlled, you might see some kind of welfare system. If not, perhaps it will work like private industry below.
If it's privatized, there will be a wealth division between those who have the means of production and those who don't - who are locked out of this high-efficiency economy. The lower 99.99% will not be able to afford the goods available in the market because they have no income. Instead they will perform whatever jobs are available.
For example, if you can't get a job, you will do whatever you can. Remember your whole neighborhood can't find work, can't pay rent. Maybe you get some seeds and plant food to survive, and in the process you have an excess. You trade this excess to your neighbors for stuff they do, such as clothing repair. Money doesn't enter the slum economy, and it doesn't get used. Instead you have a network of loans and payoffs, of gifts given and received. People live in relative squalor to today because they, as an economic group, must fulfill all of their own needs.
This may be supplemented by government assistance. It's essentially a tax on the working and owning 0.01% to make sure the 99.99% don't starve to death. But any assistance from the big capital-owners isn't going to help. For example, if someone in the slums buys a pair of shoes on credit, how will he ever pay it off if he can't find a job? And that tax may be unsustainable, if the robotic production is geared toward producing massive amounts of goods but can sell to only the 0.01%.
Production will not only have little to no market of customers (no demand, because nobody can afford the goods at that price), those customers will demand durable goods because they can't afford to buy new clothes every five years or a new washing machine every ten. The culture of obsolescence will no longer be tolerated. People will be forced to do without a washing machine, and wash clothes by hand.
Which brings up why the slum economy will still work: there are human-labor alternatives to automation. A human can grow a crop of tomatoes and mash and cook them into pasta sauce. A human can wash laundry, or repair a car, or teach someone how to play baseball. So these people will survive. But their standard of living will not be great because they're basically self-employed subsistence workers completely without working capital.
It may be cheaper to own a motorcycle, for example, than a horse. The horse eats a lot, it moves relatively slowly, it gets sick, and becomes too old to work long before the motorcycle can no longer be repaired. However you can breed the horse - when's the last time you got your bike to do that? But the main sticking point is that the horse requires only things that can be produced within the slum economy. The motorcycle requires some kind of fuel / electricity, spare parts, tires. It's not about cost and usefulness, it's about having the right kind of currency.
Would that split between a local slum economy and a global robotic economy be stable? Would the poor, as evidenced in all of recorded history, rise up and demand a redistribution of control over production? This would effectively mean that the original investment in the infrastructure is lost, and production owned in common means goods produced are available in common. It would topple every throne of the 0.01% - but likely they would stand back up still taller than everyone else.
What does a person do all day when robots produce goods and serve human needs, and they can produce art, compose music, and explain the beauty of a sunset over the mountains? When the only reason a robot lover is less enjoyable than a human one is merely that the human is human? Will people begin to attribute maximum value to handmade goods rife with imperfection and relatively expensive to produce?