Tim wades in to the debate about robots replacing low-skilled workers. I don’t think I made my case as strongly as I could have, and consequently I can’t resist responding, if only briefly.
- I didn’t react to the “cost of materials and energy” argument as forcefully as I should have. I really, really, really don’t think this is going to be an issue. Machines are much more efficient than humans — that’s sort of the whole point. It’s not at all a subtle difference, but rather a massive leap (they don’t call the first sweeping wave of automation a revolution for nothing). The price of technology continues to fall. And the energy costs we’re talking about are slight: keeping a Roomba charged takes about a third as much power as illuminating the lamp on your nightstand — and that’s if the lamp is using a compact fluorescent bulb.
- Tim’s discussion of our hardwired preference for other humans is an interesting one. But I’m not sure this is a problem that will be as resistant to engineering as he thinks. Consider Paro, the robot seal. By all accounts he works pretty well at soothing nursing home patients. Tim’s probably right that people would prefer — and pay for — a human R.N. rather than Rosie the Robot in nursing scrubs. But if a suite of relatively simple assistive technologies is cheaper than either — partially automated showers, biomonitors, pill reminders and the like — they’ll probably settle for that.
Human labor remaining competitive with robots seems to assume that one of two things will happen.
First, wages for unskilled labor could be pushed even further down. Presumably this is supposed to happen as average income continues to rise. I’m not sure how this can be considered feasible, either practically or ethically — we’re basically talking about reinventing the serf.
Second, robots could become dramatically less economical due to some sort of resource or energy shock. But given robotic efficiency (and the reality of their still-falling prices), it’s hard for me to envision this happening except in the sort of catastrophic scenario in which maintaining human employment levels would be the least of our worries.
“Wages for unskilled labor could be pushed even further down. Presumably this is supposed to happen as average income continues to rise.”
I think you’re ignoring the fact that peoples preferences change as they get wealthier. If you’re making $50,000/year, you’ll probably opt for Paro for Grandma. If you’re making $250,000, you’re likely to decide that having a human nurse is worth the extra cost. The same is true of a lot of the consumption choices made by people at the high end of the income spectrum. The difference between the middle-class version and the rich-person version of a service is often that the rich-person version is a lot more labor-intensive. Which is another way of saying it involves employing a lot more (often unskilled) workers.
I think the reason people prefer nurses isn’t just that nurses currently do the job better in a technical sense than machines. I think it’s also because people generally enjoy interacting with other human beings, and are willing to pay extra to do so. It would be perfectly feasible, and maybe even cheaper and more efficient, to build an upscale restaurant where you order on a computer screen and the food comes out of a conveyor belt. Yet no upscale restaurants do this, because flirting with the waitress is a valued part of the dining experience.
Sure, machines can be cheap and efficient. But so long as there are positive prices on resources and energy, there will be tasks where it’s just not worth it to have a machine do the job. It doesn’t cost nothing to build, program, operate, and maintain a machine. So in considering any given task, the question will be whether it’s worth going out and obtaining a machine that can do the job or just using a human. Humans aren’t necessarily all that power efficient, but they have the notable advantage of existing whether we want them for tasks or not. We don’t only build them when the marginal benefit of their manufacture outweighs the marginal cost. And so there will be times when it’s just not worth it to go to the trouble of getting a machine, particularly when the machines are only a little better at a task, rather than much better.
And the resource and energy constraints will most certainly bite when you’re trying to get a robot to do everything people currently do — build houses, police the streets, etc. And yes, people are energy hogs too, but they already exist and are going to need energy whether they’re working or not. May as well be working.
Robots doing the work of humans isn’t about replacing the people. It’s about doing things that are unsafe, undesirable, or otherwise impossible.
This isn’t about having a robot replace the seamstress, it’s about creating technology that will replace seams.
There will always be room for craftsmen, of which I’d consider high quality Nurse care. But I want my cyber-pants!
Sure, you don’t have to pay to build humans. But they have a minimum cost. You can’t just say “might as well keep people busy” — at least, not unless you’re proposing a make-work subsidy program. There are plenty of things it would be nice to have someone do, but which no one is willing to pay to have done.
And remember, we’re talking about unskilled labor. I don’t think that policing really qualifies. And for construction and sewing: robots already do a lot of that work!
You can still find human labor used for things like the shipbreaking example. But that’s because so much of the world remains desperately poor. That should be expected to change — people will expect to consume more and more per capita even as the resources necessary to build a robot remain constant.