Hunter-gatherers were displaced by a less labor-intensive occupation: farming with very primitive tools. But net employment increased -- because the number of workers needed to secure a subsistence-level food supply decreased, freeing up people to work in other fields, such as building temples out of blocks of stone.
Methods continually improved (but very slowly at first). Thousands of years later, the tools got more advanced: using a sharp-edged scythe during the harvest. Net employment increased again, in many diverse fields including supporting infrastructure such as mining ore to make those scythe blades.
Then agricultural machinery came in. Advancements were happening faster, and not coincidentally, even as agricultural employment collapsed, the number of people employed in all fields rose faster than ever. We're talking about this: as all of the old tasks became easier to handle, an even greater number of previously-unimagined new tasks popped up to take their place. There's been a transition from >90% of all workers being needed to grow food, to less than 2%; that's pretty close to ALL the old tasks no longer needing human labor. (90% of all workers were displaced by mechanization, if not by automation; the semantic distinction between those two words makes no difference to former ag workers.) It's indisputable that the new tasks employ more people than the old tasks. We're better off for the disruption.
Physical labor is over and done with soon enough.
It's been done with for a while... look around at the billions of humans employed in 2016. Hardly any of them are doing manual labor. (Jobs like operating a backhoe do not count as manual labor.) And good riddance. Unless you know someone who, say, makes hand-dipped candles for a living, everyone you know owes their job to a modern technology that was disruptive at the time it was introduced, and works in one of thousands of new fields whose existence had not been imagined 200 years ago. From the person who supervises supermarket self-checkout stations, to the person who lays fiber-optic cable. Even the job of a modern schoolteacher is sufficiently different -- utilizing instructional videos delivered over the internet, in an air-conditioned, electrically-lit facility -- to say its existence had not been imagined 200 years ago. These fields were not "visible" 200 years ago, but it would have been silly for the policymakers of the day to wring their hands about that fact. Net employment exploded, in spite of the fact that the new fields of employment were completely unforeseen, and no one would have been able to answer your demands to "again tell me" where the displaced workers will go to get employed.
If I have a company doing manufacturing, and I used to employ 50 people. I automate the whole thing, and I need to keep 5 people employed to do maintenance and oversight. Okay. How then, in your opinion does this translate to companies needing more labor, when what's happening is that 1 person working likely less than 40 hours a week is now capable of performing the labor that used to take 10 people 400 hours? For more jobs to appear to employ the 45 now unemployed people, demand needs to go up. But for demand to go up, people need to have money to spend to create the demand, and it's not clear or automatic under future scenarios that such an increase in demand will happen.
This is exactly the argument of the Luddites 200 years ago. They protested the automation of textile manufacturing; all textiles were formerly woven by hand. They lost their battle, and the results were:
- Clothing became far more affordable. Millions of people were, for the first time, able to wear more than rags.
- There was a huge net increase in employment. Former textile weavers were indeed displaced out of their old jobs, and into the new fields of employment that grew as a result of the populace having more disposable income. Good riddance to the drudgery of weaving textiles by hand.
My grandmother was a middle-class woman who never got a pedicure. She probably viewed them as a frivolous way for the idle rich to spend money. My wife -- also a middle-class woman, but with more disposable income, thanks to the generally-increasing purchasing power of middle-class incomes, thanks to automation -- gets pedicures a few times a year. The number of people employed as pedicurists has spiked, to name just one of the multitude of fields of employment that are currently exploding.
Now, what if my wife someday comes to prefer robot-administered pedicures... at that point, would I start to become alarmed about robots replacing human laborers? No, that would still not be evidence of an end to the pattern that has existed for thousands of years (and grown more obvious in recent decades); it would still not be evidence that previously-unimagined fields of employment will stop emerging. If pedicurists move on to different jobs, which don't involve handling others' feet, they will likely say "good riddance."
Some people make the mistake of thinking "more people are employed today simply because more people exist today." This field-of-dreams philosophy ("if you create people, jobs for them will come") is dead wrong: while there was a population explosion in the last century resulting in 7.4 billion people today, automation is the only reason those people not only have jobs, but are able to support themselves at an unprecedentedly high average standard of living. If the same baby boom had been attempted 3000 years ago, or 300 years ago, it would not have resulted in billions of employed humans; it simply would have resulted in a lot of dead babies.
we cannot assume that current levels of employment, which are already falling, can be maintained when the efficiency goes up across the board.
I see you're still sticking to the assertion that a world where more people are employed than ever before is a world of "falling employment."
In the future, employment can be maintained at or above current levels, by
- encouraging people who value leisure time to negotiate shorter workweeks. (For a given amount of work to be done, a two-hour workweek employs twenty times as many people as a 40-hour workweek.)
- reaping the dividend of falling birth rates in developed nations. (Falling birth rates are a bane if your Social Security system was structured as a Ponzi scheme, but a boon if you're concerned about providing jobs for everyone who wants to earn an income.)
- and most of all, recognizing that workers will be entering entirely new fields, most of which are presently unimaginable. (There will be fascinating vocational challenges for people tasked with, say, building a ground transportation system, and all necessary supporting infrastructure, between Armstrongville, Mars and Musk City.)
I guess you are saying that AIs and robots will truly outperform humans at every conceivable task, so human labor will no longer have any use as an economic input. In that case, we will well be able to afford to do some things inefficiently, and some tasks should be reserved for humans anyway. Maybe the bots should be programmed to clean up litter imperfectly, so the job of some humans could be, once a week, to pick up a single piece of litter. On a superficial level this may seem silly, but it has value. The humans must keep some small amount of "skin in the game." I will say again that a work ethic is an ennobling thing -- perhaps the ennobling thing -- and it must not be stolen from our species. Without it, people will descend into currently-unimaginable levels of brattiness. In a mild form, some people will go about spraying graffiti, just for the amusement of being followed around by a robotic cleanup crew. And who knows what the brattiness would look like in its more severe forms? There's often a lot of truth in the old aphorisms, such as the one about "idle hands..."
Just as there is value in visiting historic sites to retain some connection to what life was like in centuries past, there will be value in performing some work.
If the AIs truly exceed human intellect, the time it takes them to go from intelligent to super-intelligent will be short, and the time it takes them to go from super-intelligent to hyper-intelligent will be even shorter. For a while -- until it becomes impossible -- the favorite pastime among smart humans will be studying what the AIs are currently thinking, and attempting to comprehend the devices the AIs are currently engineering. If we're lucky, there will be scenes like this: a kindly robot approaches you and says, "please consent to this injection of dermal nanobots. They will restore pigment production, so your hair is no longer grey, and cure your psoriasis with no side effects." Because it has been years since any human has been able to comprehend the devices they engineer, you now have a choice. Trust the bot? Or, as the conspiracy theorists have been saying, is this at long last their "final solution" to get rid of the parasitic humans?
And you smile as you think back to the quaint days when the debate of employment vs. basic income seemed like our greatest concern.
P.S. Happy New Year