[First appeared in Forbes on August 7, 2018]
Amidst all the hype over AI, there is also a lot of disquieting buzz about the negative consequences of AI. These fall broadly into three camps: job loss, criminal use, and ethical issues. Here, I focus on the concerns of job loss because this is a very real issue.
Philosophically, I am a card-carrying member of the AI fan club, being involved in it for the past three decades, starting with writing expert systems in Prolog back in the mid-80s and moving on to research. I always thought—and continue to think—that outside of low-temperature physics, AI is the coolest invention to beat all inventions in the history of humanity.
So, what more is there to fear here? I won't go into scholarly statistics about the number of jobs that have already been lost to AI with many more potentially on the chopping block. I won't harp on how the transportation industry, facing acute shortages of truck drivers (surely one of the most mind-numbing jobs ever), is looking into AI "to dump the fleshbag[s] and let the trucks rumble solo down the highway."
In my latest book on AI+Blockchain, I describe how jobs that we thought must require a human touch and are therefore safe from AI aren't so immune after all. A quick example should convince you how a number of AI-based technologies can unwittingly collude to make this possible: driverless cars, robotics, payment systems, vision, and natural language understanding for a start. These form a synergistic continuum that, in the restaurant experience alone, can eliminate the parker (valet), barker (doorman), greeter, checker (coat and hat), seater, order-taker, server, sommelier, manager, entertainer, cleaner, and cashier. Let's not forget, tucked away in the kitchen, possibly robotic cooks. Let's face it, fast food burgers don't require a 5-star chef, do they?
All this is widely available on search engines for your reading pleasure (or agony, depending on your viewpoint). But what you won't see as much is the bigger revolution in jobs that I think is inevitable, whether one likes it or not, a revolution that will take place quietly at first, and then with increasing impact.
This revolution will have a much deeper impact than the mere (!) loss of jobs, devastating as job losses can be. After all, unemployment is nothing new. Anyone remember the following industries? Buggy whip makers, switchboard operators, or lift operators? How about knocker-uppers, aka wake-uppers: people whose job was to knock on the outside of your window to wake you up at a specified time in the morning; seriously, that used to be a job 100 years ago before the widespread use of the alarm clock!
Except for the most ardent Luddites, all reasonable people accept this as the price of progress and adapt in their own way. This adaptation is possible mostly because of the disruption caused by automation generally takes a long time to make a whole industry totally extinct.
Here is what I believe about AI and job loss:
Q: Will automation, specifically AI-driven automation, eliminate jobs?
A: Yes. Lots of them and in the most unexpected ways and at an unexpected pace.
Q: Will lost jobs be replaced by other jobs, just as it happened so far throughout history?
A: Only to a limited extent; there will be a massive net loss of jobs. I know many scientists and thought-leaders whom I respect a lot are predicting a huge increase in AI-related jobs to more or less compensate for the losses. I think this time they are wrong. When jobs were lost to mechanization, jobs for the mind opened up. What will happen when jobs for the mind become unnecessary or uneconomical?
Q: Should we retrain our children to compete with robots?
A: This question doesn't even make sense. Did you retrain your kids to run faster than cars? Did you retrain your kids to become stronger than tractors?
Q: Should we refocus children's education to those areas that AI cannot address?
A: Yes and no. The skills that are immune to AI—and unfortunately that is a wispy barrier—are the most impervious to mandatory formal training methods because they stem from innate passion or prodigy-like natural skills. Examples are higher math, theoretical physics, the higher levels of art and music, and so on. For these types of skills, I am reminded of what Edward Gibbon said, "But the power of instruction is seldom of much efficacy, except in those happy dispositions where it is almost superfluous."
How would you adapt to a revolution that promises, not in words but by example, to eliminate the notion of jobs itself? This is the unthinkable frontier beyond which few people are venturing. The idea that we must have jobs is so ingrained in human culture that we forget to ask why.
Let's step back for a second and look at this objectively. The reason we need jobs is so we can pay our bills, buy things, and fulfill our desires. What we really want aren't jobs but the income that jobs bring us. What we need is work (not jobs) for self-fulfillment, and therein lies an enormous difference. More on that below.
What proportion of workers can truly claim that they love their jobs? Talk is cheap. The acid test is this: If you were to win a lottery jackpot worth several millions, would you continue to work at your current job, perhaps for free? I suspect that even those who claim that they really, really love their jobs would re-examine their options if they hit the jackpot. People may continue working, but they wouldn't hanker after paying jobs, especially not jobs that are ridden with stress, abuse, harassment, and office politics.
Isn't this the promise of automation? Let's take a walk down a path that admittedly stretches our credulity. Imagine a future where most of the work is automated, thanks to AI. This would include farming, transportation of goods, personal transportation, maintenance of roads, generation of power, manufacturing, custodial services, interaction services (for example, customer service and in restaurants), medical care, elderly care, schooling, and so on.
Note that I'm not saying humans will be—or even could be—completely eliminated from the equation, only that their role will become either minimal or move up the cognitive scale; they would function as casual supervisors or auxiliary helpers when things get really tricky, and that too on very reduced work hours.
Far out? Not at all. This is already happening in isolated examples. As William Gibson said, "The future is already here - it's just not very evenly distributed."
Informed people know that this is already happening. What stretches the imagination a bit is how this affects jobs. Let's get back to the fundamentals of economics - the law of supply and demand and its effect on prices.
When AI eliminates jobs (more accurately, the need for them), there is the obvious loss of income. This means less disposable income and reduction of spending on nice-to-have goods and luxuries. Less demand forces prices to drop. If prices drop below a level where commodity margins are threatened, the company and eventually the industry, will fold.
Prices of necessary goods (generally, commodities) will continue to drop, but shrinking margins are somewhat offset by decreasing operational costs (thanks to AI-driven automation). Food prices, for example, could go down. Remember that the sinister food cartels that we fear (whether truly sinister or not) cannot keep charging seemingly high prices because consumers' inability to pay cannot sustain high prices in the long run.
Rather than outright and widespread job loss, what I think will happen is a gradual lessening of work hours, eventually moving employees from full-time to temp. But, if all things go well, the loss in pay need not pinch because there will be a corresponding decrease in costs for all types of goods.
People will slowly come to the realization that total income is less important than relative income. Indeed, income that is measured relative to one's neighbor is even less important than measuring against its economic power. Would you lose much sleep if your six-figure salary got cut in half if home prices were also slashed to a fifth, a pound of filet mignon cost $1.50, and a 10-lb bag of potatoes cost 10 cents? Would you really care if the richest person on the planet makes one hundred million times more income than you do under these conditions?
I don't think people would worry so much about the end result if all things go well. The problem is in the transition, of course. Not all things go well all the time. But the general direction will be towards less job-related work without any corresponding reduction in the quality of life (and perhaps an overall improvement in the quality of lifestyle). This progress will obviously not happen in a straight, happy path, but will meander agonizingly, causing considerable distress, backtracking, and excitement along the way.
In truth, I suspect that there will be plenty of advance notice of impending job loss. A few hundred thousand truck drivers aren't going to show up for work one morning only to be told that they have all been replaced overnight by driverless trucks. Drivers' ed instructors will not be completely blindsided to find one day that they have no students because everyone switched over to driverless cars. The entire DMV staff isn't going to be laid off on the same day because autonomous vehicles and blockchains put them out of business overnight.
There will be a transition which, despite all the hype and exuberance, will have a fairly long trajectory. But slowly, slowly, across the board, people will start discovering that they really don't have to work 60 hours a week. There will be serious proposals to reduce the work week from 40 hours to 36 hours; then to four days a week. Full-time work gives way to part-time work. Desk-bound, 8-5 work will slowly give way to flex-time work. Remote work will edge out on-site work. People will cut their working hours by 25% and find, to their pleasant surprise, that their lifestyle actually improves a bit as a result, even if they aren't making the previous level of income. Hard work will be replaced by smart work (but 'smart' perhaps not in the sense of making enormous amounts of money with little work, but in the sense of obtaining high returns on happiness and lifestyle). Financial freedom will give way to time freedom.
AI will cause enormous job losses, but that is nothing compared to the loss of meaning in traditional work and virtue of duty-bound toil. This will force people to re-examine their motivations for work. Would you get up in the morning and do something for your own sake, for your own pleasure and passion, when you don't have to do so to feed yourself?
What will you do when you don't have to?
Many people will be unpleasantly shocked to find they have no passion. For them, AI spells debilitating inertia, deepening despair, and the slide into apathy. A happy group (hopefully, more than a few) will look upon this as a golden opportunity to get rid of the shackles of work they must do so they can do work they want to do.
The AI revolution will be eclipsed by the revolution caused by decoupling jobs from work and by moving from "I work, therefore I live" to "I live, therefore I work." Sociologists will be forced to re-examine and re-formulate their models of human interaction and organization. Economists will be forced to re-think incentives and agency relationships. Politicians will be forced to invent new rhetoric for their platforms when the traditional political posturings will become moot. Schools will be forced to contend with a deschooled society (in the sense of Ivan Illich's "Deschooling Society").
Society as a whole must then grapple with the deeper social, economic, and psychological ramifications of permanent net job losses caused by AI. In consolation, the loss of jobs without (optimistically) loss of lifestyle should give us the time and freedom to think about these issues. Perhaps we can even debate the nature of intelligence as it applies to artificial intelligence.
If all this pans out as I think it will and should, this is quite a good impact of artificial intelligence on natural intelligence; we have nothing to fear but to be prepared for the transition itself!