AI agents pose an extinction level event for office work and the income it provides. White collar workers who don’t stockpile assets now are sleepwalking into poverty.
By 2030, white-collar employment will collapse. Not because of a recession. Because we’ll be obsolete.
I don’t like the idea, but not liking it doesn’t change anything. Just because I’ve spent 30+ years doubling down on human intellectual labor doesn’t mean I get to ignore what’s about to come. Neither should you.
Whistling Past the Cognitive Labor Graveyard
The engine of our obsolescence won’t be a cyclical downturn, a market correction, the looming federal deficit, or the handwringing in the news about tariffs and taxes.
It’s the relentless exponential surge of AI across every domain of cognitive work. As I explain in Fall of the Ladder Climbers, we’re not merely witnessing the automation of routine tasks, as we did with previous technological waves.
This is different. AI is getting better, faster, smarter than we are on every metric, at unprecedented speed.
On Tuesday I configured an AI agent to do internet research, create summary briefs written in my voice, and graphic design them using my brand. It took me three hours. It runs 24×7, never gets sick, costs me pennies. One year ago I would have had to hire people to do this.
CEOs of major AI providers themselves are publicly sounding the alarm – Anthropic’s CEO has warned that ~50 % of white-collar entry-level roles are on the verge of being eliminated. If the real numbers were scarier, could they really say? Or would the public backlash make it impossible? They’re walking the line between selling the dream of the technology and warning of its risks.
The critical point isn’t that your specific job title will be phased out (though it will, sooner than you think). It’s that the fundamental economic calculus that once valued human thought as a scarce, premium input is being shattered.
The entire edifice built upon the presumed superiority of human cognition in the workplace is crumbling.
This Time is Different, Because Cognition is Different
In previous technological revolutions, humans have always elevated themselves. That’s because they’ve been biological, mechanical or communication advances.
- Biological leverage (agriculture, domestication, vaccines) freed us from the vagaries of nature so we could think: the species advanced.
- Mechanical leverage (plow, train, factory) freed us from a life of back-breaking work so we could think: the species advanced.
- Communication leverage (writing, radio, Internet) sped up how quickly we could think: the species advanced.
Each revolution freeing us from something slowing us from deploying our most potent asset, our intellect.
This time is different, because the revolution has created something that will be (or is already) cognitively superior to us. And even if you disagree with that assertion, you’ll be forced to agree tomorrow – because AI is going to start recursively self-improving – which will mean a rate of change humans cannot comprehend. It’s already doubling its capability every six months.
We are hitting a “peak horse” moment. The number of horses in gainful employment peaked around 1910 and then declined sharply as cars and tractors took over. Horses were muscle labor, and when machine leverage was superior horses simply became economically obsolete.
It’s that moment, for us, for our minds. Yet, despite this stark reality, many will cling to the comforting illusion that human ingenuity will, once again, triumph and secure our position as the apex predator. These arguments aren’t just flawed; they are dangerous distractions that will leave the believer unprepared. So let me dismantle them.
We Are Not Special: The Anthropocentric Myth
All the counterarguments you’ll hear are premised on a near-religious belief that we are special. That it can’t possibly be that we no longer inhabit a privileged place in the universe. But that’s kind of like my cat. He’s sure that he’s in charge and can’t possibly understand how much smarter we are than him. The difference is we’re a tad smarter than cats. We can understand we aren’t special.
Centaurs Remain Mythical
The “Centaur” model is the idea that humans will be augmented with the tools that AI brings. It’s the analogue to “we’ve always elevated ourselves in other revolutions, so we’ll do it with AI.” Proponents have us orchestrating fleets of AIs to do our bidding, securing our place in the world above the machines.
Right now, I can use AI agents to replace mid- and low- level human workers, so perhaps today we have a centaur model. It can’t last; the ability to replace increasingly skilled humans with machines will only accelerate. Why would I have a slow, inefficient, HR-issue-riddled human working for me, orchestrating smarter, faster, cheaper AIs?
This is my cat thinking he’s orchestrating my household. Jokes aside about cats actually being in charge – nobody in their right mind would hire my cat to orchestrate me.
This isn’t just about incentives – that obviously employers will make the decision to replace people with AIs. It’s that they’ll have to. It’s their fiduciary duty to shareholders to effectively deploy capital. A company deploying a cat instead of a genius in a box can neither compete nor meet its ethical obligations.
The “New Jobs” and “Human Creativity” Myths
The WEF – and others – make the argument that even if AI will replace existing jobs, entirely new jobs will be created that we can’t even fathom today. After all, a hundred years ago software developers weren’t a thing.
And what about creativity? Some cite “AI can never replace our creativity.” Even if we leave aside that AIs are (surprisingly) creative already and getting more so, ask yourself this: do you really believe that all the hundreds of millions of office workers will be able to quickly re-tool themselves from printing out TPS reports to a career of creatively orchestrating AI?
These arguments would hold merit if the proponents could demonstrate that there’s some job that a human is uniquely qualified to do, that a machine can’t. We know we’re not the strongest, most people just aren’t that creative, and now we’re not the smartest.
The Ragged Edge of Human Replacement
It’s fair to say that everyone won’t be out of a job at the same time. The transition will be uneven. Depending on your job, you might get a little reprieve – or you could be on the chopping block in a year.
Demand-Side Preference
My wife is a pediatrician. She can’t conceive of people wanting to have an AI as a doctor. They’re not yet good enough, true. But even when they get as good it’s because she believes people want a human to interact with. I think she’s right – for now, for a little while. Today’s parents might want a human to deal with. But as One Medical has shown, they don’t always care that it’s a doctor, it could be a Nurse Practitioner.
In the medium term, the relentless cost/efficiency won’t justify a human who has spent decades training when the same medical results could be achieved with an AI paired with a human “customer service” role. What may be the nail in the coffin, however, is legal liability. When medical outcomes from AIs are consistently superior to human providers (as is increasingly the case in radiology), the standard of care for medical practice will dictate using a machine, not people – it will be malpractice to not use an AI.
In the even longer term, later generations will have been steeped in AI for every aspect of their lives and might not care if their medical advice comes from a human or an equally empathetic machine.
Regulatory Brakes and Government Inefficiency
If there’s one thing that’s guaranteed in life, it’s that government will slow things down. I don’t expect this will be any different for AI. However, governments seeking to regulate AI will face the same pressures as businesses: an “arms race” dynamic. Jurisdictions that over-regulate will lose, just as Europe has been seeing for the last decade. AI is borderless – and capital and innovation (and the tax revenue it brings to offset labor losses) are global, and will flow to environments of least resistance.
Amusingly, our bloated, inefficient, tech-backwards, wasteful government might slow down office job extinction a little. With over a third of GDP being government spending, a slice of our economy might remain in human hands for longer. A government job might become one of the last places humans will be paid for their inefficiency.
Polanyi’s Paradox
MIT’s David Autor highlights “Polanyi’s Paradox,” which is the idea that we know more than we can tell. In other words, many human tasks rely on tacit knowledge or situational adaptability that isn’t easily written as code or learned by an AI. For example, a handyman’s flexibility, a kindergarten teacher’s intuition with kids, or a business strategist’s creativity. This seems on its surface a reasonable point – but even if true, it only delays the inevitable. AI may only start with crude simulations of some professions, it’s just a speed bump to a technology doubling capability in months. There are already people getting help from AI therapists. Even with ‘intuition’ you can’t outrun exponential self-improvement paired with exponential cost decrease.
Fits and Starts
Companies don’t just fire everyone and put in AI; they test it, find errors, deal with customer backlash, etc. If an AI customer service agent frustrates callers, a business might reintroduce humans despite the AI capability being there. Again – until the simulated behavior reaches human level, for pennies.
The end of jobs as we know it will be staggered and chaotic. But nothing stops this train.
Prepping for a Post-Labor Economy
The question isn’t whether this is happening. It’s happening, and likely sooner than we expect – or want. It’s what we should do. If in the relatively near future we can’t earn based on our labor, how do we find economic security?
People have floated ideas like Universal Basic Income (UBI), essentially ‘welfare for all.’ Putting aside where governments will get the tax receipts for UBI without confiscatory rates, I’m not terribly interested in a wait-and-see strategy that will leave me in a bread line with everyone else.
For those who want to preserve their economic agency, it means a rapid shift from labor to ownership. For the last eighteen months I’ve been telling everyone “we have until 2030 to stockpile as many assets as we can” – because the world is going to get very, very bumpy.
How Will Traditional Asset Classes Fare?
In a post-labor world, many assets won’t behave like they do today. As white-collar jobs start to vanish, how each of them act as investment and stores of value will change. Let’s look at them in rough order of where most hold their assets and see if it’s a good place for us to weather the stormy years of transition.
Residential Real Estate
Roughly two-thirds of white collar workers own homes, and count it as the lion’s share of their wealth. The reality of residential real estate is that price appreciation has been driven by monetary premium, not increases in the value of the underlying assets. Houses are made of sticks and plasterboard, and that doesn’t actually get more valuable over time. Homes increased in dollar terms as a store of value against inflation, buoyed by preferential tax treatment.
What happens when white collar workers, most of whom have just have a few weeks of rainy-day cash, start losing jobs en masse? A series of mortgage defaults, and a lack of new buyers. Even if we assume some government intervention (mortgage payment reprieve, subsidies), monetary premium will deflate and this will irreparably shake confidence in the residential real estate as a way to store value.
In other words, storing a significant portion of one’s wealth in a home might be a very bad trade in the not too distant future. There will be some geographic exceptions, and ultra-high-end homes of the monied who don’t live off income may fare better, but it won’t be pretty.
Other real estate subclasses might not fare well either – I know I won’t be renting office space for my AIs. It’s possible that industrial will be less affected, robots still need a place to build things.
Equities and Market Index Funds
White collar workers hold a lot of equities in their retirement accounts. How will these perform? Companies will be able to increase margins substantially be shedding human labor. On the other hand, the demand side might crater because laid off workers don’t have money to spend. While these long-term supply and demand dynamics will eventually play out, the immediate impact of widespread layoffs is likely to be severe market shock.
Markets are already sensitive to not-that-bad news, and it’s a fair guess that significant unemployment will spook investors who will liquidate and flee to more reliable store of value asset classes. The Great Depression featured a –89% drop in the market, and we’re looking at higher unemployment without the prospect of a recovery. Obviously this time is different, and there are more ‘safeguards’ now – but still a risky bet to place in my book.
Depending on the severity of unemployment, one could wonder if governments could raid retirement piggy banks. Just last week Australia introduced a plan to bring in a new 15% tax on unrealized gains in retirement accounts with balances over $3 million. Equities, as highly regulated and centrally controlled pots of assets might be a tempting target.
Bonds
The flight to safety from equities might initially push capital towards traditional bonds, particularly government issues. However, their viability as a long-term store of wealth becomes questionable. For corporate bonds, widespread issuer defaults would be a concern. For government bonds, outright nominal default by a major sovereign issuer is less likely – but a real threat comes from inflation. If governments respond to economic collapse and mass unemployment by putting the money printer on overdrive to fund support, the resulting inflation would decimate the real value of most fixed-income assets. Even inflation-indexed bonds could face challenges if confidence in government stability itself erodes. Maybe a better bet than equities, but riddled with problems.
Gold
For centuries gold has been the “risk off” asset of choice. The OG of hard assets for storing value, and a traditional hedge against systemic risk and currency debasement. Its relative scarcity (about a 2% inflation rate as more is mined) gives predictability. It’s having a good run of late as the money printer warms up. As long as sovereigns continue to stockpile, it might not be a bad place to store wealth if you’re OK choosing between low transportability and paper gold rehypothecation risk.
Asset Classes For The Post-Labor World
Traditional asset classes – other than perhaps gold – might not be the best place to store wealth as the world economy re-orders itself. So if we’re investing in assets to weather the storm, where do we put our cash and our efforts?
Privately-Held Businesses
Businesses agile enough to arbitrage the AI transition could get a significant (but temporary), advantage. Smaller businesses that rapidly integrate AI to slash operational costs, enhance productivity beyond human limits, and innovate at machine speed will be able to unseat incumbents that can’t shed human labor fast enough. Businesses where the “last mile” or “real-world” interface still demands a physical presence have even longer to run, because autonomous robots are further off than digitizing office humans.
But let’s call a spade a spade. The average white-collar professional working a desk job won’t (and can’t) do this. Business ownership is trench warfare even in stable times, let alone during breakneck technological acceleration. Entrepreneurs might have a ball. Cubicle dwellers won’t make it. So this may be a solution for you but entrepreneurship isn’t a solution for the masses.
The real issue is that at the speed of disruption we’re facing, private businesses won’t really be places to store wealth because their lifespan might be far shorter. Ten (even two) years ago you could buy a small or medium business and have a decent bet at good cashflows for years. Soon, not necessarily. Between competition speeding up and possible demand-side collapse, the risk-reward is going to be very different soon.
Brand (Personal & Business)
Two years ago I saw a deepfake for the first time. It wasn’t very good. But I saw where the puck was going – and planted a personal brand flag on YouTube.
Digital avatars are now really, really good. Soon they’ll be perfect. I’m on record as a real human from before the age of deception. Proof of humanity will matter, because people want to trust real people. And trust is an asset.
As products get easier to replicate, “what” becomes a commodity, and the “who” becomes the differentiator. Layer on that we’re already drowning in a post-truth media environment, and it’s only going to get worse. The question we’ve always asked will have even more import in the future: “who can I actually trust?”
Whether it’s for their business or for themselves, those treating brand as an asset (rather than an afterthought) will be one step ahead.
Proprietary Data & Data Sources
AI is already chewing through the masses of data we generate every day. Even now with humans managing them, they’re extracting a surprising amount of signal. When AI agents can make near-instant decisions to analyze something, they’ll want access to even more. Our digital sawdust won’t be enough. The phrase “data is the new oil” will have renewed meaning as machines find new ways to consume, process, and transform it into useful intelligence. In an AI-driven economy, owning unique datasets or gatekeeping access to data flows will become a source of wealth and power.
Art & Collectables
Creation is becoming a commodity. Certainly digitally it’s becoming trivial to generate works of art (written and images, and soon flawless video), but soon in the physical world as well. Nearly everything new will be a commodity. In this new world, the authentically handmade, the provably scarce, and even the beautifully imperfect – the wabi-sabi – will command a premium. Not in spite of their flaws, but because of them. In a weird way, the labor theory of value may become true for art.
This dynamic is set to drive increased value for certain art and collectables. As an asset class they suffer from low liquidity, but might be a fit for those looking to store wealth with a long time horizon.
Bitcoin
Bitcoin is the digitally native store of value asset for a world run by software. It’s a good bet that the Bitcoin network will be the rails on which autonomous AIs transact without meddlesome KYC/AML restrictions. Until recently it was considered a “risk-on” asset, but it’s delaminating from correlation to equities and is a provably scarce, infinitely portable, better version of gold. It also addresses the confiscation risk of holding equities with true self-custody and portability.
Software may be eating the world, but Bitcoin is eating money. My choice as the best asymmetric bet for the coming storm. All the other post-AI asset classes require effort or skill, but buying Bitcoin is simple for all hairless apes.
Which Should I Choose?
Of course, these aren’t the only assets you can choose. But hopefully you see how to think about your options. Or just make it easy on yourself and buy Bitcoin.
One Minute to Midnight
The vast majority of office workers will wake up from their cubicle slumber one day to find themselves out of a job, with no prospect for a new one, just three weeks of rainy day cash, and a mortgage that still needs to get paid.
They’ll probably never realize that they spent years doom-scrolling their way past learning about AI, in search of the next outrage-fueled bait piece of non-news, spending like tomorrow will be the same. That they used their energy to fight culture wars from the last century, obsessed with a world that no longer exists.
I hope I’m wrong, but I’m not.
Most of them won’t wake up, won’t be saved, and hand-wringing about the morality of it all may feel right but doesn’t do anything.
The thing about exponential change is it sneaks up on you. Flat, flat, flat, a little less flat, even less flat, BOOM. We’re at the ‘even less flat’ stage now. If you wait until everyone is talking about it, you’ve waited too long, you’ll have far less runway to start saving, and assets will already be repricing.
When you get on a plane, the safety presentation always says the same thing. Put on your oxygen mask before assisting others. My mask is on, and writing this is my attempt to help you put on yours. If you’ve read this far my hope is you’ll start future-proofing yourself to live on assets, not income.
Because you won’t like the bread line.
