Life After Work
How AI, robots, and automation will redefine our relationship with work—for better or worse
AI can—and will—affect the demand for human workers. We’re quite aware of that by now. That’s a rational thought, though. In practice, unless we’ve already faced it, we feel safe. I feel safe.
Automation may drastically affect others’ lives, but we see it far from us.
This possibility probably doesn’t concern you. You aren’t going to lose your job to AI. But how are you so sure? Let me tell you that, whoever you are and whatever you work on, you’re not safe.
Because right now, no one is safe.
This is a story about the present and future of work. About how our relationship with it will irreversibly change. And what we can do—individually and collectively—to better digest and prepare for the impending transformation of our society.
It’s not a question of if, but a question of when.
Technology is a double-edged sword
This idea isn’t new or specific to AI. Since the dawn of time, technology has acted as a force that pulls civilization forward while indifferently leaving many people behind, to their fate.
And never uniformly: Vanguard and rearguard get more distant at each step of progress. Rich get richer, poor get poorer—relative to the other group.
The printing press, the combustion engine, modern machinery, and the computer… all great inventions have redefined society for the best. For whom? Many suffered the consequences then—but we’ve forgotten because we can’t feel their pain. We only care about the progress we were given.
How did workers adapt then? How many lost their jobs? How did they relocate in society? We don’t think about that because technology creates more jobs than it destroys—eventually.
Did those who suffered the advent of new tools or techniques fill the new roles? This question only matters when the “who” is me.
We, humans, are generally delusional in our self-importance: Maybe you, like me, perceive AI as greater than all preceding technological threats—something fundamentally different—because it’s a sign of our times. It is us who face the implications of progress now.
I could argue that AI is unique historically speaking because of its transversality and potential, but I’d be deforming its relative significance because of my close relationship with it—not only in knowledge but also in contemporaneity.
In its double-edged nature, AI is just another paradigm-changing technology—like the internet, electricity, or the printing press were before. It’ll create more jobs for new generations—who won’t care, as we didn’t, about the jobs it’s going to eliminate.
These are our times and the anticipation of suffering takes a whole new meaning when it’s us who will have to deal with it.
AI and robotics: 70 years of accelerated change
AI (or robotics) isn’t a concrete invention or discovery. It’s not a single point in time. More akin to physics than electricity, to biology than genome sequencing.
As a scientific field, AI entails different paradigms and inventions, none of which has made a deep-enough dent in the fabric of society to stand out by itself. In contrast, it comprises many small but compounding advances that, over time, are changing society at all levels.
That’s why both AI and robotics feel so gradual. They’re vast oceans, rising their level year after year, slowly asphyxiating everything. The printing press came around in the 15th century. Its inner workings and immediate implications were plain clear. Even if people refined it over the years, its purpose was always the same. Workers could adapt.
Adaptation requires understanding. AI is so complex, and diffuse, and ever-changing that attempts at preemptive adjustment feel vain.
Let’s see how much AI and robotics—as two intertwined forms of progress—have evolved throughout the last 70 years. And how much they’ve impacted human work.
From narrow dumbness to general intelligence
Since the early 60s, narrow-purpose, dumb robots have been mass-replacing humans. Automotive companies were pioneers—in part because car-making was dangerous and in part because of improved efficiency. They’ve come a long way to fully-automate factories.
Mass-produced robots don’t just belong to remote factories anymore. They’re at home. By 2019, there were 16 million robotic vacuum cleaners in the US alone. Also narrow and dumb—can’t replace humans by themselves—but they’re pieces of a larger puzzle: Domotics’ ultimate goal is to eliminate human household services (hi, Alexa).
Those narrow-dumb robots are the preface of a world populated by general-purpose humanoid ones. That’s what big robotics companies are betting on anyway. Agility Robotics’ Digit or Tesla’s Optimus are intended to perform “like a human” in changing—and challenging—environments.
In combination with their narrowly-designed cousins, they could soon manage a whole manufacturing plant: Inhuman factories are incoming.
Cost may not be a barrier anymore
Technological readiness isn’t the only obstacle to mass adoption—and the subsequent mass replacement. Cost is a key impediment for medium and small companies to leverage the benefits of automation.
But a new market is emerging. Companies like Formic, Locus Robotics, and Rapid Robotics lease robots, removing the otherwise extremely high up-front costs.
Polar Manufacturing, one of Formic’s clients, spends $8/hour on the robot—which is half the $15/hour minimum wage. Lower cost, higher productivity: The perfect recipe for mass lay-offs (they love euphemisms).
Will they fire their human workers? Not necessarily, but they “may not need to hire” another ever again.
Westec Plastic Corp employs three robots from Rapid Robotics. The company saves $60,000/year and doesn’t need to raise the wages of the robots—which work non-stop, and more efficiently than their human counterparts.
When costs drop, adoption accelerates. It’s expected the number of leased robots will grow from barely more than 4,000 in 2016 to 1.3 million in 2026.
Expect an unstoppable transformation in all manual labor-based sectors.
Autonomous robots will conquer every industry
Flippy 2, from Miso Robotics, isn’t intelligent nor humanoid but it’s making its way into fast-food kitchens. It’s been deployed in 100 locations in the US already. No one wanted to work flipping burgers—now no one will.
Human waiters and waitresses play a harder game: decision-making plays a bigger role in these jobs than we concede. Robots aren’t anywhere near level, but some companies, like Pudu, are already trying.
Higher-stake settings require even higher precision. By 2020, 5500 Intuitive’s da Vinci surgeon robots had been installed worldwide. Although now remotely operated by a surgeon, it’s the seed of a human-free precise, and reliable future.
Although cars aren’t yet fully autonomous, Tesla, Waymo, and Cruise’s self-driving technology will completely redefine the transportation industry: Delivery, heavy trucks, bus, and taxi drivers… millions could become jobless in the next decade.
John Deere plans to take self-driving tech to revolutionize farming. The company’s CTO recently said they’ve “now figured out how to decouple the labor from the machine … the farmer is not even strictly necessary.” Great.
These are but a few examples of a generalized trend.
A study by Daron Acemoglu (MIT) and Pascual Restrepo (Boston University) found theoretical evidence that, from 1990 onwards, industrial robots have impacted job opportunities and wages negatively. The effects will only get more acute.
White-collar workers aren’t safe either
So far, I’ve only talked about robots; physical machines for manual labor (note that robots are, strictly speaking, AI-powered in many cases, but they rarely implement neural networks to function—although that will change).
I’ve focused on blue-collar jobs because, historically speaking, the consensus was that manual labor would be first replaced by robots and AI technologies.
This has been—to the growing fear of many—a most famously failed prediction. The huge success of pure digital deep learning-based approaches has shifted the general perception.
When it only affected blue-collar workers, many thought: You saw that coming, you should have updated sooner. Well, now we’re part of them.
As OpenAI’s CEO, Sam Altman, succinctly argued last year:
AI systems can, perhaps unsurprisingly now, perform cognitive skills—both repetitive and creative—at a near-human level. The foremost examples are language models and generative AI.
GPT-3, which has been repeatedly surpassed by its successors, can enhance your writing professionally: Ads, emails, copywriting, CV generation, team management, content marketing, and note-taking.
And creatively: Dialogue, impersonation, essays, news articles, plot summaries, tweets, teaching, fiction, poetry, songs, humor, online games, board games, memes, cooking recipes, and guitar tabs.
GPT-3 powers a whole new wave of tools that have impacted the demand for writers, copywriters, editors, marketers, and administrative people… and it won’t be long until everyone has access to them.
If anyone can be a writer, what’s the value of writing well?
More recently, as you all know very well, AI art models like DALL·E, Midjourney, or Stable Diffusion have been sending shockwaves throughout the creative landscape. They can create beautiful paintings from text prompts.
In case you have forgotten, here are some examples:
That’s just images. You can also do this:
Just a week ago, Meta AI and Google Brain took another leap forward: Now text-to-video is possible.
Soon, the film and video game industries will have their movie-making and game-making AIs. Then, why would they spend millions paying hundreds of people to make a blockbuster?
The implications are huge, and artists know it. Debates about copyright, plagiarism, or work replacement won’t stop the unconstrained progress of AI.
Making a Stable Diffusion-like model is becoming increasingly cheaper as better algorithms and optimized hardware emerge every few months. Reproducing code is virtually costless. And open-source is trending among AI companies.
Imagine how this concoction will impact cognitive and creative jobs.
A new era of automation—for all?
From narrow to general. From dumb to intelligent. From physical to cognitive. Automation is slowly conquering every corner of human ability.
One argument against the supposed doomsaying about AI taking our jobs is, as I wrote above, that technology always creates more jobs than it destroys. We don’t have reasons to believe AI will be different.
Jobs lost, jobs gained, no? Well, to me this feels like a PR attempt at dismissing the inevitable cost people will have to endure:
Are those affected by automation the best suited to take newly created jobs? What if those replaced lack the required qualifications? Is it reasonable asking them to update their skillset as a solution?
One argument I hear often is that workers need to improve their employability status. As I see it, the “if you can’t get a job is because you’re not employable” mindset looks only at one side of the problem. It’s an “every man for himself” approach to a collective problem.
Don’t misunderstand me. I don’t advocate stopping technological progress. Not even for slowing it down (except maybe under specific circumstances). It’s unreasonable asking companies to not make use of new technology that could significantly reduce the costs of their business.
No, what I advocate for is human-centered progress.
First, if we assume progress always precedes law-making, regulatory organisms should get ahead—maybe by employing insiders who know what’s happening in the field. Otherwise, companies will—as they have—enjoy an “anomalous space of non-accountability.” Thankfully, that seems to be changing both in Europe and, more softly, in the US.
Second, regulation of these practices is paramount to provide a safety net to those people who won’t have the opportunity to get back into the workforce. The advancement of technology is inevitable—and crucial for the progress of society—but shouldn’t occur at the expense of people’s well-being.
Governments have to find a compromise solution between allowing companies to exploit the productivity improvements that AI and robotics provide and ensuring good working conditions for people.
For instance, by providing companies with incentives to hire human workers or establishing a universal basic income (UBI) powered by “robotaxes.” Elon Musk, famously anti-regulation, advocates for this solution: “In the future, physical work will be a choice … This is why I think long term there will need to be a universal basic income.”
Let’s get utopic for a moment and imagine governments around the globe implementing UBI. How could anyone criticize automation then? The result would be net positive: More jobs overall, fewer jobs we don’t want to do because they’re dangerous or boring, better technology, and financial solvency for everyone.
The thing is, securing income solves only the obvious side of the problem. Work provides people with money but also meaning. And that problem can’t be solved institutionally. If we get to that point, each of us will have to grow beyond our current relationship with work.
Life after work
It all comes down to this. This is the future that AI and automation promise. Is in this sense that AI and robots are fundamentally different than previous technological advances. No other invention can promise this much.
Working from home has unveiled a well-kept secret for many of us: Life is more than working at an office. Not all jobs can be done from home, but those of us who have experienced it know there’s more to win from removing commute and increased workplace freedom than there’s to lose from not going to the office.
Pandemic-induced full-remote working has been the first wave of change. Once automation touches everything, we’ll face another wave that will radically redefine our relationship with work.
Even if we, as a society, find a way to co-live with robots and AI in an economically-stable synergy, we’ll still face another, more abstract problem. Because we spend so much time working, we define who we are to a considerable degree through our work.
Working isn't just a matter of money—that’s the basic necessity. There’s a higher-level necessity of self-realization and fulfillment: Work gives life meaning.
We’ll have to re-learn that work isn’t just about putting in the hours and earning a paycheck at the end of the month. We’ve been taught that to enjoy life, we have to work a fixed set of hours and produce high-quality results. If we don’t, a pang of powerful guilt takes over. The roots of this feeling grow deep—so much that even realizing it’s there feels like an accomplishment.
After automation sets in—and if a safety net is in place—we’ll be set free to search for personal and professional projects to work on, untied from salaries, and also redefine our understanding of what makes us deserving of a happy and peaceful life.
Most people would still work even if it wasn’t necessary. There would be more entrepreneurs embarking on ambitious endeavors. More artists, writers, musicians, and other creative professionals, who, liberated from financial needs, wouldn’t perceive generative AI as threatening.
There would be more philosophers and scientists asking and answering questions about the world. And more humanitarians willing to help our society improve.
What we won’t do anymore is suffer to pay our debts. We won’t ever sacrifice relationships or personal wellness to earn a few more dollars.
Work will have a new meaning, decoupled from the word necessity. At least, that’s what I hope for.
An issue I think about some is mental health therapy and the job of mental health counsellor. On the one hand it has features that cry out for automation: it is hard to find a counsellor, appointments are often inconvenient -- sometimes highly so -- in time, space, and money. On the other hand the resistance to the idea of a robot therapist by potential patients and clients is extreme. In my experience people regard the idea as about a ridiculous a notion as there is -- maybe on a level with robot Rabbis. But I admit that if I knew enough about the process I might see the logic in this very high degree of skepticism. But I can't quite give up hope. I am surrounded by people who need therapy and can't get/afford it.
I definitely agree that automation is a good thing, and that if we pair it with socialized medicine/healthcare and possibly universal basic income, hopefully it will only mean that humans can engage in more meaningful work. I’m excited to explore this further in future posts!