Machine learning has been driving a new wave of automation in the past decade. The speed of this wave is up for debate, but its direction and strength are beyond doubt — for it is driven by good old economic interest.
It always helps to remember:
Corporations swarm towards profit surely as maggots to a corpse.
And unimaginable profit lies just beyond the mechanical horizon. The lauded principles of the Free Market move companies to pay workers the absolute minimum, so the idea of a worker that needs no wages is a utopian dream for every CEO with their souls oriented towards the bottom line.
Many economists are skeptical of the coming robotic revolution. They point out how humans have always been automating our work (the water wheel, for example), and we simply make new jobs to fill out the slack.
This is good, I will concede, so long as the displaced workers can find meaningful work afterwards. I don’t imagine this to be the case in the future.
We Live in the Beginning
With the rapid development of computer vision through convolutional neural nets, we’re rapidly approaching the age self-driving cars. To put that in perspective, there’s ~3.5 million truckers in the US, nearly a million ride-service drivers and perhaps half a million bus drivers. That’s about 3% of the 165 million workforce who may have to look elsewhere when companies decide it’s much cheaper to buy and maintain a driverless vehicle.
How many jobs are actually at risk in general? Figures vary: 20 million US factory jobs by 2030, 25% of US jobs, 15–30% of workers displaced by 2030. There’s myriad factors influencing the speed of the wave.
Notably, the sources I linked were written pre-Covid. Visions of the “six-feet office” have pushed many firms to downsize and turn roles remote. Many companies are investing heavily in automation infrastructure instead of rehiring.
This, too, makes economic sense: The grand economic machine has been grievously disrupted, and nobody knows when things will return to “normal”. The drawbacks of automating your workforce — losing out on face to face human interaction among them — have been rendered null by the pandemic. Machines can’t get sick.
The wave of automation moves unevenly. Simpler, repetitive tasks lend themselves to automation much more than abstract, cognitive work. Maybe you’ve seen the kiosks that take your order at fast food joints; we’ve begun to implement burger-flipping bots as well.
This means that certain industries will be displaced before others:
The more predictable the task, the more likely it will be automated in the near future. To consider how much your job is at risk, ask yourself:
“How often am I faced with an unknown variable or situation? How often am I surprised?”.
These elements of surprise make it more difficult to find a viable machine learning solution to a given challenge.
This visualization from McKinsey offers a good overview of “how susceptible is this industry/role to being replaced”. It’s interesting to note the differences between industries — data collection is more variable in healthcare than agriculture — but keep in mind that this is calculated with “currently demonstrated technology” in mind. Every step forward in AI changes these likelihoods in strange ways.
The automation of these jobs poses a difficult challenge: How do we keep these displaced workers productive? We’ll be losing chunks of various industries piece by piece, in smaller or larger groups. Where can they go?
This is why people often refer to coding as “the new literacy”. There exists this idea that there’s enough room for tech (and to a lesser extent, all of STEM) to grow to encompass the increasingly-displaced workforce. Whether this is the correct or best solution is another matter entirely, but I would wager it’s more realistic than the government redistributing significant resources in this interim period.
We’ve identified the wave of automation and established its growing momentum and uneven effects. Now I want you to consider the “endgame” of all of this.
What’s the purpose of machine learning? To make our lives easier, one would hope, but more practically it’s to increase productivity.
The economically-driven goal of AI is to increase work efficiency — to reduce the amount we have to work, or enable us to achieve more with the same effort.
As more and more jobs are automated, is there enough room in the tech economy to accommodate our displaced workers? Perhaps.
But more then that: if technology continues to develop at this pace (or any Moore’s trajectory), is that new work even necessary?
Now we’re delving into the realm of post-scarcity philosophy. Far in the future, certainly, but the necessary conditions will inevitably be met as the world continues to progress.
Consider the hypothetical: If we suddenly automate half the jobs in the world tomorrow, we’re suddenly generating the same GDP/productivity with none of the labor (and paying a fraction of the price in machine maintenance). What do we do with the new “slack” made possible through automation?
There’s two paths to take:
- Companies reap huge profits from no-wage automated labor. Displaced workers must ‘up-skill’ and find new jobs
- The “slack” wealth is redistributed through UBI or other programs. People spend their lives beyond ‘productivity’
The former is much more likely to happen, at least for a while. Many will feel left behind, but the God of Capitalism is more than happy to leave them behind — it’s their fault for not changing with the times, of course.
The second is a more benevolent dream, but I believe it may be a predictable destination. Universal Basic Income is a controversial idea, but the pandemic is speeding it up. The problem has always been the argument of inflation or relative pricing and spending; I would argue that “slack” productivity from automation will solve this issue.
Which path you believe in is largely based on your answer to the question:
Is machine made to serve man?
Or is man made to serve machine?
The Nice Way to Say this is “Friction”
The government moves as its corporate donors decide, so we’ll surely experience the first path in the coming future. They’ll call it something pleasant like ‘productive shifts’ or ‘gainful friction’.
Robot took your job? Buck up, kiddo. Go give those bootstraps a pull.
And then reality begins to catch up:
If wealth begins to concentrate even further among the corporate elite, then there’s less money in the pockets of consumers. Consumers begin to spend less.
Suddenly nobody’s consuming all these new automaton-created products. Profits plummet and wealthy people start paying attention.
I haven’t even mentioned the enormous societal unrest that unemployment and displacement will cause. Look at the past few months in America:
Long-brewing issues of social and racial justice have boiled over into active protests. But why now? These problems have existed for ages.
The lockdown changed individual incentives. Plenty of people suddenly lost jobs, or can’t go in to work in person, or lost hope of employment for the foreseeable future.
Suddenly people aren’t obligated by economic interest (ex. “the freedom to starve”) to work obediently as part of the Machine, and they find themselves with the freedom to express their displeasure towards said Machine.
When you run the world by distracting people with bread and circuses, you’d damn well better keep giving them bread.
The Nature of Wellbeing
Is is just to dictate that all men must work merely to exist? We may move beyond the question of “just” in a few decades.
Proponents of “From each according to his ability, to each according to his needs” never quite measured up, because they relied on humans to push a system (inherently less productive than capitalism) beyond scarcity.
But if automation gives us that slack — if robots have the capability to “free” us from labor — what should we do?
Much has been written on man deriving his meaning through labor. There is certainly truth in finding fulfillment through creation and hard-won accomplishment, but I believe this belief has been skewed by the fact that man has always had to labor merely to exist.
The few elites who could afford leisure, who had the freedom to pursue “higher” crafts, certainly enjoyed their adventures. There is another argument to be made of debauchery born from excess — look towards the fall of the Roman empire — that idle hands are surely the devil’s playthings.
But for all of history, most humans haven’t been able to decide this for themselves. The vast majority of people labor as needed to stay alive, and do their best to find meaning in their given work & some leisure during fleeting free hours.
One wonders if the idea of “you can only find meaning through work” has survived because it helps citizens tolerate situations that otherwise seem meaningless, because many despair of wasting their lives on activities they despise.
I envision a future where man is freed from the necessity of labor, where technological advancements deliver us from the cult of capital and launch humanity into an unexplored void of being:
When we don’t have to do anything, what do we do?
Whatever we find interesting.
I could criticize my own argument as the fantasies of a lazy dreamer, but it is difficult to imagine automation will slow down. A predictable obstacle is that the economy does its best to keep people “productive” when that productivity is no longer truly needed (imagine tech & all white-collar industries becoming as saturated as the podcasts).
We may have a hard time detaching from the notion that people must work to exist. It is ingrained in our cultural psyche, because we’ve always needed to work to exist. What sort of creature are we when completely free? I’m terribly interested in finding out.
Will we be happy? Perhaps, perhaps not. That warrants another investigation of economic psychology and the productization of expectation. But people are generally happier doing things they’re interested in, so one would hope freedom brings joy.
Either way, automation will remove the largest obstacle (excuse?) holding us back from finding well-being, and it will be up to every individual to find their meaning in the void. How exciting.