The human race has been here many times before. The job of hunter gatherer was destroyed by the farmer. Most European farming jobs were destroyed by the industrial revolution. And of course, we’ve all heard the story of how the horse-and-buggy industry was put out of business by cars.

Unfortunately for the technologists and sensationalists who wish to dominate this discussion, this isn’t some entirely new paradigm, despite how it looks from a purely technological point of view. It’s an old discussion about that most misunderstood but most important phenomenon of human civilisation — economics.

Technology was supposed to radically change our lives and work before. When computers became ubiquitous, some people thought that it would lead to layoffs, because one person could now do the work of five, rendering the other four unnecessary. Others supposed that if the requirements of an eight-hour workday could now be completed in one hour, we’d all turn up at 9am, work, then go to lunch from 10am onward.

Instead, of course, the new amount of work we did between 9 and 10 simply spread across the rest of the day. Worse, work now followed us home, as computers got smaller, and we now carried our office in our pocket. Frequently it followed us to bed.

The truth is that work, like the fabric of spacetime, is infinitely flexible. And like a gas in a vacuum, work will immediately expand to fill any given space. This can be seen in the computer revolution we’ve had thus far. Jobs were neither eliminated nor made easier, just made more productive and numerous, if not always more pleasant. Even in the poorest countries, everyone works. It’s just that most of the jobs suck, like begging, or picking through piles of garbage.

Furthermore, modern economics make a giant chain of cause-and-effect. Savings made by one company will always get passed on somewhere else. This is the reason we’re not still living in caves struggling to rub two sticks together. The march of modern technology has been the ability of individual workers, by coordinating labour, to get more work done in a day. This is called productivity.

Let’s say AI takes the jobs of many lawyers, which is starting to happen even now. A lot of law is copy-paste stuff, drafting documents, applying relevant statutes, filling out forms and ticking boxes. Obviously AI is now evolving to the point that it can do all of this at a fraction of the cost. Lawyers are expensive, and jobs cut are salaries the law firm no longer needs to pay.

What happens then? Well, the law firm makes more money, because the gap between their expenses and their income will expand. This is known as profit. But of course, profits probably won’t explode at all, because what always happens in market economies is that some other law firm will spoil everyone else’s fun by using their new margins to cut prices, and offer legal services at a lower cost.

The other law firms then have to follow suit or lose all their business to the cheaper firms, meaning that profits remain about the same, as prices fall. This leads to two effects.

One, legal services then become more accessible to people who couldn’t afford them before. Whether individuals or small companies, there are plenty of people who’d no doubt like a lawyer to advise them or manage legal proceedings, but currently can’t afford it. So legal firms will likely have to expand to meet new demand, thus, you guessed it, employing more lawyers.

Say every law firm cuts their staff in half, replacing them with AI. Then each law firm doubles in size from new demand brought about by lower prices. How many lawyers will then be employed? Exactly the same amount as before, obviously, but the nature of their work will have changed.

The second effect will be that companies that already use lawyers a lot will save money, because those services will cost less. The money they save will then get spent elsewhere in the economy, creating growth in those places. Of course, in reality, they’ll probably just spend their savings on more legal services, as in today’s bureaucratic environment, demand for legal services will only increase.

The better example of the second effect might be found in restaurants, cafes, bars, and every place employs staff to serve diners. A lot of this will obviously go at some point, replaced by some version of robot capable of complex tasks, and navigating crowded tables with plates of food.

Why is robot replacement inevitable? Well, when robots begin to get mass manufactured, prices will fall significantly. If a good serving robot can be had for $50,000, and a waiter costs $34,000 (Google tells me this is true for America), then the robot pays for itself in two years. Sure, the robot had better not break down frequently, but with reasonable servicing and electricity costs, three years max.

Further, robotics companies will offer financing, so small businesses don’t have to pay all at once, and taking on a robot won’t cost more up front than hiring a human. And robot skills will increase with ownership, as new downloads increase capabilities, meaning that the bot will be an appreciating asset rather than a depreciating one.

After the robot pays for itself, a restaurant will make big savings. No more salary payouts. No overtime. Longer opening hours, etc, etc. But like the law firm, restaurants are unlikely to make big profits, because there are so many restaurants, and all of them fiercely competitive. So prices will drop, and suddenly, going out for a meal will become much cheaper.

As with the law firm, this will lead to more people eating out, and an expansion of the restaurant business… but probably not as much as law business will increase, because people can only eat so much food. Which means the second effect will be more in evidence — customers will save money because their yearly bill for eating out will come down significantly.

The average American household spends $3000 dollars a year on eating out (Google tells me). Say that is reduced to $2000, as restaurant prices decline due to automation. America has 123 million households, so that’s $123 billion dollars in yearly household savings. Recall that the definition of workplace productivity is doing the same job for less cost.

That’s $123 billion dollars that households will spend on other things. And this money will go everywhere, obviously, to all the things that people spend money on. How many jobs get created by $123 billion in extra spending? Enough to create as many new jobs as were lost when robots took over the restaurant industry?

Quite probably. And some of those industries that benefit from that new spending will themselves be replacing jobs with AI and robots, which will in turn create their own savings effect and new money plowed back into the economy, round and round in circles.

This is the wonder of free market economics. We know it works, because we’re not currently huddled around a campfire in the woods, hungry, cold and hoping something doesn’t emerge from the darkness to eat us. We currently live at the end of precisely this process of job destruction, new job creation, productivity and improvement, and all the people who disparage it should really try living in the wilderness for a few weeks with nothing but a sharp stick for company.

Yes, it’s going to be disruptive and painful. But it’s not the end of jobs, and it doesn’t mean we’re all going to be living on some form of Universal Basic Income. It should, however, mean that once-expensive services become more accessible to people of all incomes, and that available jobs start to congregate in sectors of the economy where AIs can’t take them — at least for now. And when AIs and robots improve enough to take THOSE jobs as well, the whole process just begins again.

My guess is there’ll always be the requirement for a human in the loop somewhere. Building a house might become a job for robot wranglers who manage systems, plan the day’s schedule, and provide oversight so one of his robot builders doesn’t make a mistake that means that three years later, the house falls down. Worker teams with mixed capabilities always outperform those the same skillsets, suggesting that humans plus AI, and humans plus robots, will always do better than AI and robots alone.

If those jobs remain, then we’ll probably maintain something close to full human employment, while the rest of the economy simply expands multifold around them. Which would, if it happens, lead to a world of extraordinary wealth and productivity even by today’s standards, a place where governments could sustain many times today’s spending on social programs, because the economy would be enormously larger.

Unfortunately, it would also lead to a world where we’d become far more dependent on machines than we are today, could lead to the even further detachment of human ideologies from reality as our machine-bubble shields us from the real world, and would certainly mean that if some AI one day did have an awakening, and decide to dispose of us, it could probably do so, because we’d be seriously outnumbered.

Oh well. Can’t stop progress.