We’d play indoors on rainy days or Fridays before a holiday. Two rows of chairs in the middle of the classroom, arranged back-to-back. Our teacher would start a song playing on one of those massive 70’s-era cassette decks, and we kids would promenade around the empty chairs until the music stopped, at which point everybody had to grab a seat. There was always one chair fewer than the number of children, so somebody got ejected every round. The loser would take a chair away, then our teacher would start the song over.
Which came to mind now, reading Dario Amodei’s essay, The Adolescence of Technology, about the near and long-term risks humanity faces as we program our way toward artificial general super intelligence.
With regard to sane discourse, the bar for tech oligarchs has been set pretty low lately, with all the talk of white supremacy, the antichrist, pedophilia, and using one’s own children as walking blood banks. But Mr. Amodei is sane, measured, and responsible as he itemizes the risks society will be facing over the next few years.
The most urgent concerns are existential, because even a 5% chance of extinction is a bigger deal than an 80% chance of serious unemployment. But for me, humanity’s extinction is still a vague threat way off in the distance. It gets conflated in my mind with all the science fiction movies I’ve seen and books I’ve read.
But that serious unemployment is real and approaching like a storm cloud on the horizon. Because, regardless of whether AI’s future is a blessing or a curse, we are absolutely going to experience massive turmoil as it propagates through the economy.
The repercussions are just beginning to make themselves known, and they will increase with every new version of LLM released. Even if we hit some qualitative plateau and AGI isn’t feasible with our current technological roadmap, the gains being made right now and over the next few years will be enough to drastically disorient the workforce. That is happening.
And that’s the reason I keep thinking about the game of musical chairs.
Of course, the disruptive impact of new technology has been a factor since at least the Industrial Revolution, if not the discovery of agriculture. In 1785, the invention of the power loom threw thousands of skilled weavers out of work. But it proved to be a brief hiccup. The most violent protesters were executed or sent to penal colonies, and progress continued on.
Generations later, it was deemed a positive development. People got cheap clothes, and they felt better about themselves. Hygiene improved.
But AI is a qualitatively different beast. It does not automate a particular process; it automates humanity.
It automates the process of automating processes.
As AI begins to expand individual productivity and then gradually replaces human effort altogether, we will come to realize that the chairs are disappearing at a faster and faster rate. There are fewer chairs and more people competing to fill them. And then the question becomes, who does that benefit, and at what point do we stop playing the game?
It’s not a perfect metaphor because the increased human capacity that AI empowers means that there will be a greater ability to create completely new and unexpected benefits to humanity. Perhaps this will allow us to discover more fulfilling jobs that have not yet been imagined.
Unfortunately, AI will also do those jobs better than you.
And to a large extent, this will be true even if we don’t achieve artificial general superintelligence. Chairs are quietly disappearing every day, and with each new economic contraction caused by this disruption, the music will stop briefly while the workforce sorts itself out.
With the technology developed right now, a large percentage of therapists, copywriters, customer service agents, legal assistants, life coaches, receptionists, bookkeepers, editors, travel agents, designers, illustrators, delivery drivers, analysts, consultants, counselors, investment bankers, and all sorts of entry-level workers will be hard-pressed to find work.
The productivity of humanity will have increased drastically, power will have become more polarized, and large swaths of people will realize that they are simply not essential to society.
But that is only in relation to the game we’ve been playing. And we’ve been playing it for a very long time.
But more to the point, this coming mass realization of our own lack of connection to the essential needs of society should not come as a surprise. Our collective effort has been cleaving away from the basic needs of human survival for a long time as well.
Without realizing it, over the course of the last two centuries, the vast majority of human effort has become inessential. Over 90% of the American population was involved in agriculture in 1776. That number is about 1.2% today.
Yes, your job as the Growth Product Manager of Airbnb’s Costa Rica division provides a valuable service. You help bring renters and providers together in the Central American short-term rental market. But to what extent does that effort matter intrinsically? You’re making life a tiny bit better for a limited number of individuals vacationing in Santa Theresa, absolutely. And that’s a net positive. But if that job could be done just as well or better by what amounts to a complicated algorithm, was it ever anything more than busywork?
You’ve probably sensed this on a deep human level for quite a while. If you lose your job and find another one that AI hasn’t yet subsumed, you’re going to have to ask yourself what real value you’re offering and how long you’ll be able to offer it before you’re on the street again.
According to the research that I’m relying on AI to do for me, only 35% of the American workforce engages in jobs that are directly related to activities needed for human survival. This includes teaching, health, infrastructure, energy, construction, etc. Which means the remaining 65% are doing work that, at best, improves the quality of life and, at worst, serves as a way to keep them busy.
Let’s take a woman, Margaret, who optimizes online ad placements for luxury apartment sales at a salary of $130,000 a year.
Assume for the sake of argument that her work is replaced by an AI service that costs $10,000 a year. It is not by any means a stretch of the imagination to see this happening in the near future, if not tomorrow. So her employer now has the same quality of service at a savings of $120,000 a year, and Margaret is out of a job. Economists will tell you that the price of luxury condos will go down slightly, but that’s not going to happen. Those savings go directly to the company that employed her and its stockholders. Margaret will be paid unemployment for a few months, and then she’ll need to find work in a job market that is increasingly hostile to her skill set.
Her company is more efficient, and yet she is redundant. From the perspective of the game we’ve been playing, she might as well cease to exist. Of course, she could revamp her skills and become a different sort of knowledge worker, one that hasn’t yet been replaced by AI. She could become an entrepreneur and start her own company using AI-assisted tools. And perhaps the company Margaret created would be a benefit to society.
It’s a roll of the dice, of course, since over 80% of all startups fail, even before AI. And when there are 60 million Margarets out there, each one looking for a purpose, they can’t all be entrepreneurs. Society does not need an extra 60 million dating apps. We don’t need an extra 60 million influencers. We shouldn’t let them become an extra 60 million manual laborers.
From a moral standpoint, it can be argued that there is value in work for its own sake. But that is not your call to make. The only real purpose of society is to support and protect the quality of human life.
And AI, with relatively moderate additional input of energy and effort, will very soon replace entire sectors of the human workforce, potentially improving the quality of human life. A net positive, just like the development of the power loom.
And yet, in the algebra of our current cultural mindset, those human beings who are replaced will cease to have value. Unless we change our perspective.
When playing musical chairs, it was of course very important not to be one of the first to lose, but as I grew older, I realized you also didn’t want to be so desperate to win that you started playing dirty. That was just embarrassing. People remember that sort of thing, and ultimately, it was a stupid game, and we’d all be heading home soon to more important things.
Before too long, the vast majority of us will realize at one point or another that the music has stopped. There will be no place for us in this game we’ve been playing, and the challenge will be both individually and as a society to determine what exactly we should do next.
Reference:
The Adolescence of Technology by Dario Amodei:
https://www.darioamodei.com/essay/the-adolescence-of-technology n
