By Adam Dorr

“AI is not going to replace human beings. It will never replace human beings.”

“AI is just a productivity tool. It increases the productivity of the worker.”

“AI will create more jobs than it destroys.”

Leaders in government and industry are trying desperately to reassure the public – and themselves – that AI and robots won’t upset the status quo. But no matter how hard we try, wishing for something won’t make it true...

The painful truth is that AI and robots will take our jobs. At the same time, the rise of AI and robots itself will create new jobs. But AI and robots will take those jobs too.

By 2045, there will be virtually nothing a human can do that a machine cannot do better for a tiny fraction of the cost. And this will still prove true (thank goodness) even if individual machines such as humanoid robots aren’t each empowered with fully sapient artificial general intelligence.

Humanoid robots today are the most expensive and least capable they will ever be, and simple arithmetic shows that the marginal cost of robotic labor will be vastly lower and thus overwhelmingly competitive relative to human labor. A robot that has a lifetime cost of $10,000, works 22 hours per day, and lasts 5 years would have an hourly marginal cost of just 25 cents. And when robots are building all the robots, they will cost a lot less than $10,000.

The marginal cost of labor will plummet toward zero as adoption of humanoid robots powered by increasingly capable AI explodes across virtually every industry worldwide. Humans simply will not be able to compete.

Let’s quickly dispel some myths that run adjacent to this painful truth:

Myth: We are still many years or decades away from AI advanced enough to replace most human labor.

While it is true that today’s AI is not intelligent enough to disrupt all human labor, the last three years of spectacular advancement following the “chatGPT moment” in November 2022 should leave no room for doubt that we’re not talking about science fiction that lies decades or centuries away. Machines that can think are here, and their capabilities expanding day by day with no end in sight.

Myth: There are uniquely human jobs no machine will ever be able to do.

We tend to seriously underestimate both how capable robots will become, and how often people will prefer a robot to a human. We are already seeing counterintuitive examples, such as a large fraction of patients preferring to speak to AI therapists rather than human ones. This trend will continue, and will encompass even quintessentially “human” occupations, such as actors and sex workers. And even if jobs persist for which the primary qualification is being human, there won’t be anywhere near enough of them needed to perpetuate the global labor market as we know it. Even if there do continue to be a handful of roles occupied exclusively by humans, like politicians and diplomats and coaches, we won’t need 4 billion of them.

Myth: Humans and robots working together will be more effective than robots alone.

There was a brief period when the best chess players were top grandmasters assisted by insights from chess programs. But it didn’t last. IBM’s Deep Blue beat Gary Kasparov, the world chess champion, in 1997. By 2015, the chess program Stockfish could defeat the strongest human-computer partnerships 100% of the time. From that point onward, humans only got in the way. Human-robot partnerships for performing labor will follow exactly the same pattern.

Ironically, chess is now booming as a sport, and the top few dozen human players are actually earning more now than in the past. But this is easy to misinterpret. We didn’t have tens of millions of people earning a living playing chess before Deep Blue and Stockfish came along. And digging ditches, sewing garments, mopping floors, and running cash registers are not going to turn into popular sports.

The point is that even if robot-human combos are more productive than robots alone for a few years at the beginning of the labor disruption, as was the case for chess programs, that won’t last for long. And so it should offer no reassurance about job security for humans.

Myth: Making AI datacenters and robots is expensive and slow, so that will give us plenty of time to adapt.

Incumbent industries always underestimate the exponential first phase of the new technology’s adoption S-curve. Horse-and-buggy manufacturers doubted automobile adoption would be swift, because in the first few years there were no factories, no mechanics, no fueling stations, no paved roads, no traffic lights, the cars themselves were expensive and unreliable, and nobody knew how to drive. The 19^th Century heating and lighting industries had the same doubts about electricity. Kodak had the same doubts about digital cameras. And until recently, the fossil fuel industry had the same doubts about clean energy and electric vehicles. Disruptions don’t take 50 or 100 years. They take 15-20 years. Sometimes even faster if demand for the high-value new product or service is large. And in the case of humanoid robots, demand is likely to be somewhere between cars and smartphones – tens to hundreds of millions of units annually at a minimum.

Myth: Even if the machines do take our jobs, it will happen slowly over many decades.

Horses didn’t lose their jobs to cars slowly over many decades. In most regions, the disruption of horses was complete within 15 years of the introduction of the automobile. And those horses didn’t end up finding other jobs. They ended up as dog food. We won’t share that fate, of course, Terminator scenarios notwithstanding. But the point is that holding out hope that the 4 million American drivers put out of work by autonomous trucks and robotaxis will make a clean, easy pivot into new occupations like software engineer and social media influencer with just a little bit of retraining is dangerously delusional.

Looking Forward

With those myths out of the way, we need to turn to the meat of the question, which I’ve written about before with Tony Seba and Bradd Libby: what does the roadmap look like for the disruption of human labor by AI and robotics, and how do we safely navigate from our current system to a future of superabundance with minimum chaos, collateral damage, and casualties along the way?

We at RethinkX don’t have all the answers. Nobody does. But we can help frame the discussion about the disruption of labor by AI and robotics within realistic boundaries based on knowledge of past technology disruptions.

A Fleeting Grace Period

To start, it is important to understand that there will indeed be a fleeting interregnum when AI and robotics appear deceptively non-competitive with human labor, for two key reasons:

1. There is enormous latent demand for labor throughout the global economy that is too dirty, dangerous, demeaning, or low paying for humans to do today.

2. AI and robots will indeed briefly act as productivity enhancers for people in existing jobs.

While true, and worthy of celebration, this condition will not persist for long. Within a few short years, all that latent demand will be saturated, and meanwhile AI will be rapidly following the chess program trajectory to become so capable that humans only get in the way.

Nevertheless, the existence of all that latent demand needing to be filled will give us a fleeting grace period to prepare for a soft landing, meaning a stable and just transformation across society in response to the disruption of labor.

The era of complementarity between labor and capital is coming to a close. “Work” will soon become something that only machines do. That means we need to begin rethinking economics itself right now because fundamental notions like scarcity and exogenous total factor productivity will no longer hold.

It is absolutely crucial that we not become complacent, and instead recognize the interim period for what it is: a brief and lucky opportunity to plan ahead for the tsunami of technological employment that is inexorably approaching.

This is why it is so important to face the painful truth, and not fall for the “nothing to worry about” line now being shouted from the rooftops across Washington DC and Silicon Valley.

So What Should We Do?

There is only one place to start: we need to talk about this painful truth honestly.

The robots are coming. They’re coming for everyone’s jobs. The train has already left the station, and we’re all onboard. Nothing can stop the disruption of labor now.

Without honest and thoughtful decision-making among leadership in every domain, and very likely a complete rethinking of the basic social contract across society itself, the destabilization caused by the disruption of labor could well be catastrophic.

The one option that is absolutely not viable is business as usual. It would be a huge error, for example, to ban AI or humanoid robots in an effort to preserve jobs and prop up incumbent industries and economies. We’ve tried to ban disruptive technology before, and it never works – it only backfires.

In this case, it would lead to a vicious cycle of diminishing competitiveness, prolonged scarcity, economic stagnation, and ultimately poverty, civil unrest, and even conflict and war. The world’s wealthy, industrial economies already host huge numbers of “bullshit jobs” (a formal academic term!), mostly hidden in the bureaucracy of governments, large corporations, and public institutions like the health and education system, thanks to the enormous productivity boost that modern technology has already given us. The United States is currently in the throes of dealing with aspects of this in the Federal government. It would be a deeply dystopian outcome if we clung so hard to outmoded economics and the notion of “jobs” that by the 2040s, the majority of people across the economy were stuck doing bullshit ones instead of the much more sensible option of retiring amidst the shared luxury of technology-driven superabundance.

Not Just Good News, the Greatest News…

The disruption – and end – of human labor has the potential to be fantastically liberating. It could quite literally be the best thing to ever happen to humanity. Societies that choose wisely today could start to look like something out of utopian science fiction by the 2040s. But only if we directly face the painful truth and have an honest conversation about what’s coming.

AI and robots will take our jobs. And it will be the most amazing thing to ever happen to our civilization, because it will mean we can all retire in luxury to spend our time (including working) on whatever we find meaningful. But the ride from here to there will be a lot less bumpy if we are honest and start planning for it immediately, instead of wasting the next decade burying our heads in the sand pretending the future isn’t coming...