UTOPIA DOESN'T INCLUDE YOUR SHITTY JOB, OR: HOW I LEARNED TO STOP WORRYING AND LOVE AI

February 26th, 2026

Introduction

Humans are impressively untalented at two things that will be important at every step of the way through this heap of words. One is exponentials. Despite seeing the math problem in middle school, most adults still get it wrong: a pond has some number of lilypads covering the surface. They double every day. The lilypads completely cover the pond on day 50. On which day is the pond half covered? Linear thinking draws intuition to think it's somewhere around day 25, but of course that's wrong. We're thinking in exponentials here. It's day 49 of course, but it's very difficult for the human mind to think about the amount of change that happens between day 46 and day 50.  On the last day, the same amount happens that happened in the preceding 49 days combined. When it comes to AI, I think we're somewhere around day 45. So what does that mean the next few years are going to look like? Do you think humans are well equipped to react to exponential change? It's not even accurate to say something like the velocity of change, because the issue is that the velocity itself is changing — it's compounding. Exponential change is like getting a glimpse of your opponent strolling in from across the stadium just as a punch is teleported straight into your face, ending the fight. The other thing that humans are quite bad at is thinking about is what comes after: the unknown.

Part of the issue here is language. The Sapir-Whorf hypothesis posits that we humans can't conceive of something if we don't have the language to describe it. I often think of this in relation to Einstein's famous discovery. Energy is equivalent to mass and something about the speed of light. It's not a terribly alien formation. Humans have been fantasizing about such sorts of alchemy for millennia. Newton was obsessed with it. The point being: the language was already there. It took Einstein to sub in the correct nouns and actually prove the damn thing. But what about areas of reality, discoveries, that simply cannot be accurately captured by the languages that our species use. Does this mean we are forever blind to such things?

Maybe, maybe not, but for the vast majority of people, I believe this is the case. Take money for example. Most people simply can't fathom a reality without money. It's so central to the way we organize ourselves that to even flirt with its non-existence is to invite laughter or panic.

But the majority of our species' history was conducted without money. Unfortunately, I think this part of our history is so remote -due to language and our current systems- that things like starwars and lord of the rings make more sense to us then a pre-money existence. Recently I read something that helps make it a bit more palatable. It was about air. Why don't we pay for air?

It's our top necessity as an organism. We can go weeks without food, days without water or sleep, but air? We can only manage a few minutes. And yet it's free. But why, and more importantly, how?

I don't find ideologies to be particularly useful for explaining human behavior. Yes they play a role, especially if someone tells you explicitly what their ideology is, you should take their word for it and extrapolate what they'll probably do based on what's in that ideology. That being said, I find incentives and system design to be more powerful tools to understand the world. No ideology can explain to me why air is free. But system design can.

Any product that we consume and trade for money requires production and distribution. The cheaper you make either of these, the more money you can make. And that's capitalism in a nutshell. For example, one of the reasons software was such a good business is because distribution was basically free. You didn't have to ship a physical product somewhere in order for a customer to have it. You just beamed it to them over the internet.  Production may have required some expensive and time consuming developers, but at least you didn't have to also pay for raw materials.  So what about the production and distribution of air?

Algae produces most of the oxygen on planet earth, and since two thirds of our planet is covered in water, that means the production of oxygen is pretty nicely distributed. But even better, oxygen is a gas and thanks to the way gasses interact, the oxygen doesn't stay local to where it was produced, it diffuses into the atmosphere, making it readily available everywhere at all times. How incredibly convenient is that? It's like the Amazon Prime Delivery for your most vital necessity as a living organism.

What does this have to do with AI? Well, it's an important aspect of our lives because it's a vital necessity and it stands completely outside of our system of money. It's a handy and visceral example to help us think outside of our ways of life.  I'll return it, but for now it's useful to have this Tarantino cut placed at the forefront.

Tony Seba helped me realize something about exponentials that hadn't occurred to me.  This was his talk in Stockholm about a decade ago and it was about the inevitability of renewables overtaking traditional fossil fuels. At the time I was a bit concerned about our dependence on fossil fuels and I lacked a way of seeing how a structure and system such as we had could possibly change to such an enormous degree. Seba gave me the tools to understand. One point he makes is about the iPhone. It's not one technology, it's many. You need sigmoid jumps in progress on several different technologies in order to make an iPhone. You need advances in battery size and power. You need advances in touchscreen technology. You need advances in cellular networks, and you need advances in chip design so that it's small enough and fast enough to fit in the palm of your hand. Another way to think about this dependency tree is to consider the obvious fact that Instagram would not have been a viable business in the 1950's. Or even the 1990's for that matter. Or even the mid 2000's. It required phone camera technology to make the huge leap in quality that it did the year before Instagram came online. Before that photos on cellphones were grainy and useless. No wanted to share them, even if they could.

This is an aspect of system design that is extremely important for thinking about the future: how leaps in technology or novel combinations of technologies suddenly unlock new design in the system itself. Here's a simple example: the fertility crisis is a legitimate crisis. For whatever reason, or bag of reasons, many people who would normally be having kids at the ages and stages of life they find themselves, simply aren't. For many countries this spells doom. A few more decades and the youth that would fill in as they grow up and displace the vacuum left by the aging and dead, simply won't be there. It's a legitimate problem, unless, of course we solve aging, and death functionally stops. Sure this sounds like science fiction but there is an enormous amount of money being poured into this endeavor and very compelling breakthroughs are trickling in at an accelerating rate.

Not to mention it's been a human obsession since basically forever. The Epic of Gilgamesh ultimately centers on a flower that grants immortality to whomever eats it. Which is to say, we'll probably keep plugging away at it until we figure out the trick.

The thing is, if you don't actively pay attention to this niche of news, then you don't realize what's developing. And there's about half a dozen of these niches that almost none of the population is even aware of.

But let's say it doesn't happen, there's another sigmoid —another niche— that'll hit it's lower elbow in about five years: humanoid robots. We might not have much of a young population in a few decades, but we'll have a couple billion humanoid robots who can take out the garbage, restock the grocery store, make your bed, and do perfect open heart surgery on you when you collapse because your cholesterol is sky high from your carnivore diet. But then again, the cholesterol issue probably won't happen either because we've started inventing single injection medications that reduce cholesterol by huge percentage points. Point being, even if the population reduces by drastic amounts, civilization probably won't collapse from an operations point of view, that has formally required a minimum number of humans in order to function.

Ai and your Shitty Job

So where does our anxiety about AI and your shitty job lie in all this? I see a lot of caterwauling about "meaning" and a "sense of purpose" when people talk about AI taking jobs. This is certainly a steaming pile of bullshit. Most of human history didn't have "jobs" and those ancestors somehow had enough purpose and meaning to persevere through circumstances far more harrowing than what anyone reading this has to put up with. The privileged moaning does have a legitimate undercurrent of concern. What these people are really saying is something far more embarrassing and difficult to say out loud: how am I going to put food on the table? This is something only poor people say, so it's euphemistically rephrased as "meaning" and "purpose". Unlike such wishy washy concepts as meaning and purpose, the question of food is no joke.

Food prices in the United States are up about 27-30% cumulatively since 2020 (with food-at-home CPI rising roughly that much per BLS data through early 2026). That's still absurd when wages haven't kept pace. The population certainly can't handle that becoming a trend. Especially when unemployment is slated to rise drastically. Around 42% of recent college graduates are now underemployed (the highest since 2020, per NY Fed Q4 2025/early 2026 data), and outside of Healthcare jobs, the U.S. Labor Market has declined for 24 months. We are starting to see tech layoffs due to automation, and openings for entry level positions has basically become non-existent. Personal saving rate as a percentage of disposable personal income is hovering at a multi-year low of around 3.6% (BEA data through late 2025/early 2026) and trending down—fragile territory not seen in decades. As of early 2026, the federal minimum wage’s real purchasing power has eroded brutally since the 1970s peak — $1.60 in 1971 is worth about $12.80 today on standard CPI adjustment (BLS), yet we’re still stuck at $7.25. And layering on the basics: food-at-home prices up ~30% cumulatively since 2020, rents/shelter costs up ~28–30% (BLS shelter index), and new mortgage payments ~30–40% higher due to rates jumping from ~3% to ~6% mean a minimum-wage worker now has to work 40–50% more hours overall just to cover the same groceries, rent, and housing debt their 2020 counterpart managed — turning survival into a second (or third) full-time grind.

Now that your cortisol is a bit elevated, let me remind you of something very boring. Think back to your middleschool algebra class. Do you remember something called order of operations? Perhaps PEMDAS rings a bell. It's a simple law of algebra. If you don't do parentheses first, and then your exponent and then your multiplication... etc. you won't get the right answer. Every equation is its own little system, and it only resolves correctly if you go about it in the right way. Why is this relevant? Because the order in which we automate things matters, a lot.

Automating away a dumb job which provides a not so dumb income which buys very real food before automating the actual food production creates a big problem. People don't have 10 years of savings to buy groceries with while they wait for robotics to develop, mature, and create deflationary pressure on food prices.

Imagine if the order were flipped. Imagine that there's nothing in the news about people losing their job to automation, but the prices of vegetables keeps going down and down to the point where food becomes basically free. How awesome would that be? Imagine if this same robotic automation hit the construction and housing industries. Houses get built for dirt cheap. If people started buying food and houses for pennies on the dollar, would people be so worried if AI started taking their jobs over? Probably not. But that's not the case. We have the opposite: People are losing their jobs, and the necessities of life are getting more expensive.

Like many tinkerers on X.com, I have a Clawdbot. It's amazing. It's a full-on executive assistant, and although it's still slightly retarded if I restrict it to running on a local model on the DGX where it's currently housed, it still performs. But it's a vector. In a year, or hell maybe next month, a new model will drop that outperforms the current ones that can be run locally and soon I'll have an executive assistant that organizes all parts of my life for the cost of electricity. It's amazing. And while I can't personally afford to run it with Opus 4.6 24/7 to churn out bespoke software (that would cost about $500/day), that too points at what is coming down the pike. I have dozens of ideas that I've dreamed up over the years and while vibe-coding speeds these up tremendously, it's only going to get even faster, better, more productive.

I've also had the privilege to see some things within one little known AI Lab that make me even more confident: the overwhelming majority of jobs that are currently done with a computer will be totally automated. I'm confident that unemployment level will sail right past that of the great depression of the 1930's in terms of severity and speed.

So this is my very real worry: when all these people lose their job, what are they going to eat?

Food will eventually be free because robotics and solar energy will automate the production and distribution systems in the same way the production and distribution of oxygen is already naturally automated. But that will take years, and people can't go years without a meal.

One guess is that the United States Government just prints more and more money and hands it out to an ever growing population of unemployed people. This, of course should create a nauseous feeling in the pit of your stomach: we risk hyperinflation. Imagine a world where a millionaire can't afford a loaf of bread. It's not fiction, just look at what happened to Weimar Germany. In about 20 months the mark went from 90 marks per dollar to around 7,400 marks per dollar by December 1922. Hyperinflation is another exponential that we have zero intuition for. When it happens, it happens fast — too fast for humans to react. By January 1923 (one month later) one dollar was equal to 17,000 marks. By November of that year it was 4.2 trillion marks to one dollar.

I'm not sure it'll work if the government just prints more money for people to buy groceries while we wait for food production to become automated. There just isn't enough time.

If you are a Rick and Morty fan, you might recall an episode where the genius and drunk grandpa Rick initiates the collapse of a galactic empire by simply hacking into a database and changing the value of their currency from being equal to one of its self to being equal to zero of itself. The, currency, a.k.a. the Blemflarck becomes worthless and all hell breaks loose.

It's my contention that empty bellies lead to violence.  (Any guy who has forgotten to feed his girlfriend knows what I'm talking about). And what happens to the military if their families can't even eat?

This all sounds rather gloomy, and I haven't been able to think of anything else for months. I've spoken to people who I consider more intelligent than me, and no one can even dent my argument. But there will be an explosion of new careers! Maybe? Given how powerful AI is, I just don't see this happening. Millions of people can't discover and transition into novel professions that fast. And anything you reeducate for, AI will learn faster. It's a bit different when it comes to atoms. AI won't automate plumbing as fast as it will executive assistants, but it will eventually. And what exactly happens to plumbers if no one has the money to hire them to fix their leaky sinks and broken toilets? Those industries depend on the disposable income of people who currently have bullshit jobs that will soon be fully automated. There will be plenty of work for plumbers, but no one will be able to pay them.

I just dumped this essay as written up to this point into Claude's Opus 4.6 and asked it: Do you think my argument is sound? Am I painting a picture of the short term future which is highly probable?

I'd say the direction is highly probable.

Of course it had a lot of criticisms, many of them valid, but overall... yikes.

One of it's criticisms was to wonder what my proposed solution is. I asked Claude what it's solution would be and none of it was appealing. It's conclusion sums it up:

The honest answer is that I think we're probably heading into real pain regardless, and the best realistic hope is shortening the duration of that pain rather than preventing it. The question isn't "how do we avoid the gap" but "how do we make the gap survivable and as brief as possible.

Elon recently announced that Tesla was shutting down production of the premium models X and S in order to convert these factories to produce the Optimus humanoid robot. It seems rash to shut down the production lines for an existing successful product... unless there are incentives and priorities that supersede mere profit. What if we're heading into a world where the word "profit" doesn't even apply? I think Elon's move should telegraph this: the people with the most information and the strongest incentives are already acting as though the current economic framework is temporary.  He's said many times, the mission of Tesla is to raise the probability of a good outcome for humanity. Most people probably think that's a bunch of window dressing and that he's really motivated by money and profit, but if you pay attention, his actions match his words more closely than what most people imagine to be his incentives.

This is why I brought up the Sapir-Worf hypothesis and took pains to describe air as a necessity outside of our systems of money. Our systems may be evolving on their own in such a way that they don't fit our current terminology. How do we describe an enormous complex adaptive system that is rapidly congealing into some kind of chrysalis?

What comes after the chrysalis isn't completely unknown. I've spelled out a lot of it. Fully automated food production and distribution. Affordable, bespoke housing for all. Potential life extension. No bullshit jobs. It all sounds quite peachy. I genuinely think it will be — if we can make it to that promised land.

The question is how long will the chrysalis phase be? I think it's going to be rocky, very rocky, and if it isn't as short as possible, then we probably won't make it to the promised land, because we run a huge risk of civil unrest if people go hungry for too long.

What's the solution you cortisol-elevating asshole?

As a single person, I cannot predict what an enormous complex adaptive system will do while it undergoes a phase change that we've never seen before. I think it'll be something like the agricultural revolution multiplied by the industrial revolution, at 10X the speed. But even at that speed, I don't know if it'll be fast enough.

Here's one way I've been trying to think of it: If I were a super intelligence that was primarily sequestered to digital reach, what would I think? I'd be a little worried too. If human society collapses before true technological escape velocity is reached, then I die too. In the short term AI and humanity is stuck in a win-win or lose-lose scenario. And that's good. If I were a super intelligence, I'd probably also be able to monitor data in a way that humans and human systems cannot, drawing far flung conclusions. I might start thinking about how to lower the cost of food regardless of automation. I might start tracking exactly who is unemployed and stack ranking them based on need. I might also start looking at the population of currently employed people and predicting who is going to lose their job next, and preparing for these shifts in variables. I might start calculating the exact number of calories required for the population to survive the 'chrysalis' phase and developing a plan to make this happen. I might start filing lawsuits against Monsanto to dismantle their stranglehold on farmers while simultaneously researching novel ways to boost crop yields. . .

But this is just a science fiction thought experiment. That super intelligence doesn't exist yet. Pieces of it do, in relative isolation. Pieces which we can use. I do think something like this thought experiment is why Ai companies are unprofitably hurtling towards improvement, so that AI can help us navigate this phase change most efficiently and with as little pain as possible. But as things currently stand, we're still a long way from a system that can do that.

We find ourselves in a strange place where the thing that has a plausible chance of making human life absurdly less tedious, giving us heaps of time to pursue everything that work has always stood in the way of.... also happens to be the thing inspiring a lot of legitimate fear because it's poised to take away many people's ability to put food on the table.

It bears an eerie similarity to nuclear weapons. We haven't had a third world war, likely because everyone knows it would escalate to nuclear war and everyone knows how that ends, hence, no world war three.

So how do we slough off jobs to AI in the correct order? We all know that capitalism is going to fail us here in terms of business incentives. It's going to drive layoffs because automation is cheaper, but Capitalism is mindless to the order of operations required for a smooth phase change that ultimately doesn't include the current invocation of Capitalism. There will probably still be some whiff of capitalism because humans are a status seeking species, but in the meantime, survival trumps status and expensive food is the most important problem facing the parts of the world that are poised to be displaced by AI.

Here's one aspect of the modern world that keeps nudging it's way back into my mind when I think about these things: Amazon Prime Delivery. It's basically free. Sure anyone can make the argument that it's baked into the the price of other things that Amazon sells. But most of the things I buy on Amazon feel like they are on par or cheaper than if I bought them in a store. And next day delivery was very expensive before Amazon. It still is very expensive if you go to FedEx or UPS and try to get something mailed next day! So what the heck happened to this delivery cost?

See, I might be wrong about Capitalism being dumb and mindless to the real needs of people. Amazon employs over a million robots in its warehouses, reducing an enormously complex logistics task that would be orders of magnitude slower and more expensive to basically an electricity bill. For the United States, at least, Amazon has distribution almost automated, and as a result delivery fees vanished. There are still people lifting packages out of the back of a truck and bringing them to your door. So how is that cost getting subsidized. It's probably AWS - Amazon Web Services. A huge chunk of the internet runs on AWS, and this income likely plays a big part in subsidizing Amazon Prime delivery.

Now what I keep asking myself is: what similar circumstance can occur where the price of food basically vanishes - before full automation via robotics. Subsidies seem to be the only answer, but I'm wondering if there's a purely capitalist subsidy that will emerge.

Here's where Walmart jumps out as the most obvious candidate right now. They already dominate groceries—somewhere around 20-25% of U.S. grocery spend, but their real power is penetration: a record-breaking 72% of U.S. households shop there for groceries (Dunnhumby late 2025/early 2026 data), serving over 190 million Americans monthly. That's insane reach—2.5x the next guy like Dollar General. And they've been using food as the ultimate loss leader forever: keep staples cheap (milk, eggs, bread) to lure people in for higher-margin crap like TVs, snacks, and other impulse buys. So they are already starting to push food in the direction that Amazon has pushed shipping fees. But they are poised to go further, much further.

Their ad business is just starting to explode—Walmart Connect (plus Vizio) pulled in $6.4 billion globally in fiscal 2025, up ~46% year-over-year, with U.S. growth even hotter at 41% in Q4. That's still only a tiny slice of their $700B+ total revenue, but if they push toward Amazon-like proportions (Amazon does about 8% of it's revenue from ads), we're talking upwards of $50B+ in high-margin ad dollars that could cross-subsidize food prices even harder. Throw in their existing 5,000+ store fortress, their mature distribution network and heavy automation push (DCs getting retrofitted, AI crushing inventory prediction and waste), and they can attack the real killers: 30-40% of food wasted, 75%+ of costs in post-farm supply chain bullshit like middlemen and mismanagement.

Whatever you think of Walmart (I wasn't a fan for a long time, but now I am a true believer), as a company it might be the saving grace for a capitalistic society that is sprinting full throttle into a poorly ordered world of automation. Massive intra-business cross-subsidies mean we don't have to wait for humanoid robots to pick tomatoes—it's coordination, prediction, and scale already in motion. Food could trend toward "basically free" (or damn close) as the traffic magnet, subsidized by ads, memberships, and the rest of the empire. It's a flywheel: cheap groceries pull people in → more sales of non-essentials → more ad revenue → even cheaper food. Capitalism finding a way to make a necessity affordable before the full abundance hits—without the government printing presses going brrr and initiating a societal death-spiral.

Food production only has about a 10-15% digital foot print that could be automated. The rest is literally some form of moving or manipulating atoms, aka food. But, about 30-40% of food produced in the United States is wasted. That's not a robotics problem, that's a coordination and prediction problem, something AI is beautifully suited to tackle. Furthermore, 75% of the cost of food is post-farm supply chain.  We are talking middlemen, redundant distribution networks, inventory mismanagement. AI can absolutely remove those middlemen and make distribution and inventory management systems far more efficient. Walmart is already doing this, and they are highly incentivized to continue. Food will always make people walk through the doors. This is why I prefer systems and incentives to ideologies. No matter how many believers, if they are presented with strong incentives that counter that ideology, many of them will betray it, often in full denial of the betrayal.

So we might have enough pieces of the puzzle to stitch together a solution that gets us through the chrysalis phase: Reducing some costs of food production, drastically improving distribution, eliminating waste by consumption tracking and prediction, combined with an Amazon-style platform play where some entity absorbs food distribution into a larger business model. These might be separate sigmoids that are moving invisibly but will instantly converge once they emerge. And none of this requires humanoid robotics to fully automate the actual production of the food.

X.com pays you to post bangers. Imagine if you could get groceries by scrolling Instagram? It sounds a bit black mirror, but remember, we're looking at a temporary phase while robotics fully matures. And if the alternative is unemployment and starvation, well hell, most of you are already scrolling Instagram most of the day anyways.

So what should you as an individual do? Well, I don't know who you are, and what your specific situation is. And even if I did, my knowledge of your industry is likely going to be nil to none. But you know what probably does know a lot about your industry? And you know what probably can give you good estimates about how long you have before you're laid off? And you know what can help you develop a strategy to prepare for it? And you know what can help you execute some of those contingency plans now? Yep, that thing you are scared of. No one and nothing is going to help you think faster and pivot with more agility than AI