The 'Great Dislocation' will be more painful than anyone realizes

The AI community has dubbed the coming AI and robot job apocalypse as the "Great Dislocation" and it's going to hurt more than you realize. Here's why.

Creative destruction is one of the favorite terms of futurists and capitalists alike. It’s the process by which innovation destroys old paradigms and introduces new systems. The tractor destroyed our need for oxen and all their associated upkeep, like farriers (hoof maintenance). Fast forward a few decades, and we’ve destroyed plenty more jobs like telephone operators, elevator attendants, and typists.

In all cases, we’ve created a net gain in jobs.

This is my favorite graph to demonstrate creative destruction.

We destroyed farm jobs and created cubicle jobs.

Whether you believe we’re heading for Post-Labor Economics—a world where the marginal utility of human labor drops to zero—or not is immaterial. We’re inevitably heading towards a huge disruption in jobs: the one-two punch of AI and robotics.

I’ve been crunching numbers with the help of ChatGPT o3 (which is very smart and you should use it instead of being a hipster about it) and here’s what we’ve come up with:

FIRST, we’ve been looking at the rise of CUA (computer-using agents). These are AI agents like OpenAI’s ‘Operator’ and Anthropic’s ‘Computer User.’ These two products are the first generation of LLM-powered agents that can pretend to use a computer in the same what you or I do. This is the first shot across the bow of what I call “KVM jobs.”

KVM, in techno-parlance, stands for keyboard, video, mouse. Therefore, a “KVM job” is any job that you can do with (you guessed it) just a keyboard, video, and mouse. Maybe throw in a USB webcam and headset for good measure. Within a couple years, we should expect the underlying performance of LLMs rise, combined with software integration, that makes CUA like Operator good enough to start taking KVM jobs.

Here’s my projected adoption rate of CUA based upon historical adoption rates of SaaS and other enterprise software. At the end of the day, CUA is just a line item like Office 365, and will likely be packaged in ELA soon.

To understand what a KVM job looks like—pretty much anything you can do remotely. If you WFH, you have a KVM job.

You might be wondering “how did you get to those numbers above, Dave?”

The short version is this: We looked at adoption curves of recent game-changing technologies such as smartphones, cloud virtualization, and SaaS platforms like Office 365. By and large, all these technologies had an adoption S-curve of around 7 to 8 years. This “seven year S-curve” seems to be a magic number for quite a few reasons.

If you compare early chatbots like ChatGPT to the “Palm Pilot” era of experimentation, then we’re just a couple years away from CUA’s breakout “iPhone moment” at which point quality and penetrate will accelerate, and inside 7 or 8 years, we’ll reach >90% Fortune 500 saturation.

It would be silly to assume this won’t dislocate jobs.

There’s an important difference between job dislocation and labor substitution. Job dislocation is when one job is destroyed, but the work moves elsewhere. With job dislocation, you can upskill or reskill and find another job. It’s just a matter of opportunity and retraining. Labor substitution, on the other hand, is when work is just outright replaced by a machine.

The savvy readers might be thinking “Aha! But what about the lump of labor fallacy!”

The lump of labor fallacy is the erroneous belief that there’s a finite amount of labor to be done, and that you can just automate it all away, then everyone can live in a Fully Automated Luxury Space Communism utopia. Personally, I don’t see any reason that AI and robots won’t soon be better, faster, cheaper, and safer than humans at all economically meaningful tasks, but even if that’s not true, it’s indisputable that AI and robots (the latest form of automation) will dislocate many jobs.

SECOND: Here’s why this is all going to be more painful than you realize.

As CUA ramp up, so too will robots be ramping up. As these machines encroach on more and more human labor, the pool of job openings will shrink across many domains. Meanwhile, as layoffs point, the supply of job-seeking workers will rise. More workers, fewer jobs? Supply and demand shows that this will dramatically depress wages. Plenty of desperate workers means more political anger, lower wages, and an economic death spiral of declining aggregate demand.

Guess how fast the robots will take off?

It looks like humanoid robots will be able to take all blue collar and skilled labor jobs by around 2041, only a couple years after CUA destroy all KVM white collar jobs. Yay.

The above data is based, again, on tons of historical data, such as the ramp-up of Internet penetration, innovation cycles of the automobile, Wright’s Law, and more. The TLDR is that we expect humanoid robotic production to pick up pretty quickly by around 2028. Right now we’re still in the “prototyping an MVP” phase. Humanoid robots are already shuffling boxes around Amazon warehouses and even spot-welding BMWs. But those are produced only in the thousands, and not particularly flexible.

Even so, the economics of humanoid robots are clear: even at $45k per unit, they are worthwhile to replace many jobs. But they will probably get down to $15k per unit very quickly, at which point more jobs are economically viable for disruption.

Let me state this as clearly and plainly as possible:

We’re staring down the barrel of a double-barrel shotgun of blue collar and white collar job attrition unlike anything we’ve ever seen, and we’ve got less than a decade to brace for impact.

One term that I’ve heard to describe this is economic compaction. The more colloquial term you’ve probably heard is “the hollowing out of the middle class.” It’s going to be even worse this time.

Economic compaction refers to a phenomenon in labor markets where workers with higher skills or qualifications take jobs that require lower qualifications than they possess, often due to limited opportunities matching their skill level. This leads to a “compacting” effect in the labor market where higher-skilled workers displace lower-skilled workers in positions that traditionally required fewer qualifications.

We’ve already seen this happening over the last couple decades. A Millennial graduates with a Master’s degree but ends up flipping burgers anyways. High-skilled worker taking a low-skilled job.

High intelligence developers are going to lose their jobs, reskill to become plumbers, mechanics, and electricians, and dislocate all but the most skilled and experienced skilled tradesmen.

“Shit rolls downhill” as they say.

The “economic death spiral” goes like this: Jobs are destroyed » Wages decline » Aggregate demand declines » More jobs are laid off due to declining demand » Rinse and repeat.

It’s basically the Great Depression on steroids, but instead of being caused by banking leverage, it’s caused by technological deflation. And it’s all coming too fast for the population (or government) to adapt. Even worse, most people don’t believe it’s coming.

THIRD: I do have a small silver lining. There are a few jobs that are going to be untouched by all this. I break them down into three categories:

1. High liability jobs, such as doctors, lawyers, politicians, and judges. These are also sometimes called statutory jobs. These are jobs where a human is legally required for one reason or another—licensure, insurance, signatures, etc.

2. High authenticity jobs, such as musicians, athletes, actors, fashionistas, therapists, coaches, and such. Even chess players and Twitch streamers count. These are jobs where humans just prefer other humans.

3. High trust jobs, based on relationships, like diplomats, negotiators, relationship managers, account execs, and also personal coaches and therapists. Humans are social animals and much of our world depends upon complex negotiation and relationship management.

In most developed economies, just over 20% of jobs require some kind of license, bonding, insurance, or board certification. Your company’s comptroller won’t be replaced by AI until the laws change dramatically, but they will use AI and robots extensively. Likewise, your human doctor will have a ton of AI and robotic helpers, but it will still ultimately be a human writing the prescription (for now). Insurance companies might actually push to force a change.

Either way, about 20% of the workforce just cannot be automated away for systemic reasons. Another good chunk of the economy is also permanent based on human preferences for human voices, faces, and personalities. Embodied presence is not a guarantee—robots are coming after all.

What do you think? Is your job safe? Are you ready to pivot? Do you think businesses will keep you on when their competitors are laying off droves of employees? Do you think the government will adapt and prepare in time?

Buckle up tight.