26 Comments

"What’s most impressive is that Claude can do all that while being the best human being you’ve ever worked with. It’s polite and encouraging, and always excited about whatever you want to do. At the same time, it pushes back against and talks me down from my bad ideas and unworkable schemes." -- This is a mental parlor trick of anthropomorphism. It was designed that way on purpose. You prefer straightforward, unproblematic engagement with a simulacrum because human relationships and interactions are messy and non deterministic. But there is nothing human about a being that never emphatically disagrees with you, challenges you based on what it really knows about your experience and values system, or calls you out in ways that really matter. And there is nothing human about a parrot that has no original thought that can only regurgitate the average value of whatever human ideas it was trained on. This can lead us down a dark path to putting way too much trust into a machine whose main defining feature is a propensity to "hallucinate" falsehoods and reinforce whatever alternative facts and biases were put there by its creators.

If companies stop hiring entry level developers, when their senior devs quit or retire they will have no talent pipeline and will be forced to hire uninitiated code prompters with no idea what they're doing. It'll be like the Starship Enterprise being piloted solely by flunked Starfleet cadets. Someone tell these Idiot CEOs that software engineering has always been an apprentice industry.

Your solution is only unthinkable to the people who hold the purse strings. We should be there already. The past 80 years of capitalism have resulted in exponential year over year growth, yet we work more hours now than ever. Corporations are mostly run by authoritarian assholes who believe their soul is redeemed by a tiny bit of tax evasion disguised as philanthropy. They will not save us.

Expand full comment

An error in your thinking might be that if knowledge workers lose their jobs, their demand goes away. What these AIs do only has value in an economy where there is demand for their work product. A possible outcome of introduction of AI would be a shrinking of the economy. Much of what AI starts to do would become stuff we don't need anymore because we are in a deep depression do to all the laid-off knowledge workers.

Expand full comment

You write "We need to reconfigure our economy to pay employees what they currently earn — in exchange for half the work. "

Are you referring to ALL workers or just the affected workers? If the former, that is exactly what FDR did in his initial efforts to build the widely-shared prosperity of postwar era.

https://mikealexander.substack.com/p/how-the-new-dealers-gained-the-ability

If you are suggesting only the affected workers, coders and knowledge workers in general (i.e. they get a doubling of their wage while the others get nothing) that would likely lead to more Trumps or revolution.

A key feature that I believe is necessary is for AI to be taxes on its productivity at the rate you would tax the labor that task previously took, or more simply, higher corporate, capital gains, dividends, rent and other investment taxes.

If you did all that the expected result would be rising wages for people who do physical jobs and stagnant wages for knowledge workers. You would expect to see a massive shift to trades, driving their wage growth down, while the wages of unskilled workers would rise.

It would be the return of the "loser-friendly" economy we used to have (working on a post on this) and a more human economy.

The stumbling block is the "bro-ligarchs" who will own this AI tech and who are going to want to use it to achieve a form of neo-feudalism as Elon Musk may even now be in the early stages of doing in preparation for this eventuality.

Expand full comment

Yeah the ideal transition isn’t some dramatic thing where all the desk jobs vanish overnight and just as suddenly people start getting compensated fairly for grueling physical labor.

Expand full comment

"Even ardently pro-UBI policy makers gave up on UBI after a few small experiments proved it impractical and undesirable."

Can you share more details? My understanding is that UBI works too well and policymakers are afraid of its success.

Expand full comment

This article is appreciated — I’ve been thinking a lot about this.

As a young g person I see the impacts of this on my own job and the employment landscape more generally. I follow Graeber in thinking that the IT revolution has *already* eliminated a lot of work, which has been filled with more BS administrative tasks, email, meetings, etc. The basic incentives are already pushing even ‘high skill’ jobs towards simplification, standardization, idiot proofing, etc. I work in a highly technical field, yet the instruments I work on have been progressively modernized. What once required a highly trained technician with a steady hand can now be done by anyone with a degree and a few hours training — honestly an attentive high school student could do most of my job. The job market has already been shifting from a ‘ladder’ model, (where organizations invest in long term talent, train them, provide employees with promotion opportunities and benefits) to a ‘stepping stone’ model (where jobs provide less training and serve as resume padding to help get your next job at a different company). It’s subtle but skill and expertise are already being devalued.

That’s *before* AI. With AI a C coding student like myself can design powerful automated workflows. People who don’t code have been able to hide from this reality. This is going to change everything.

I have many concerns.

One concern that isn’t mentioned enough is that nothing will change as dramatically as we think. This is actually bad. It imagines a future where the incentives to maintain the employment status quo persist, but the reality of what people do at work trends towards total bullshit. You still have the serfs cleaning the floors overnight, you still have the professional class showing up for the 9-5. It’s just that they increasingly act to officiate AI. It’s the ultimate triumph of bullshit jobs. Huge wasted time doing low value activities at work because the incentives to maintain the status quo don’t go away.

I do think it’s more likely there is an acute labor crisis coming, however. AI is close to a tipping point. The cost savings and business advantages are going to be too large. Just think about call centers. The first big company to employ AI agents will save big on staff AND customers will get perfect customer service with no wait every time. No more confusion. No more talking to a guy in India who has no idea what’s going on. It’s a no brainer. That’s just one example that could easily roll out in the next 1-2 years (*easily*).

I just wonder if you aren’t thinking big enough. This won’t just come for “knowledge workers”. I mentioned call centers. I saw a driverless car when visiting Austin TX. Not on a special track, not with a person behind the wheel just in case. Driving on the roads with nobody in the car. Robotics has advanced a lot since 1900 and the incentive to automate will go up if everyone is earning double the hourly rate. In my opinion that’s by far the biggest gap in your plan. Also, it imagines little room for humans to specialize and do amazing things with what skill they do have. At the same time, as someone in the comments mentioned, this whole economic system relies on consumer spending. If 70% of people get fired there won’t be an economy to get rich in.

You mentioned the problems with UBI but only briefly. I just wonder if that’s at least part of the solution. Moreover, I increasingly question our dominant economic paradigms. People look at the Soviet Union and see evidence for the superiority of our capitalist system. I see it as evidence that whole societies can be captured by flawed economic ideologies. Maybe our economic system has inherent flaws too, maybe it won’t cope with these changes and things that seem impossible will be possible. Change is the only constant.

Expand full comment

I know virtually nothing about computing, but here is the rub from an ignorant perspective: AI requires vast amounts of water and energy to run. And being human made, will never be perfect, or human, even if they can mimic and analyze and decipher faster than humans. The data is all input by fallible and creative humans who have both the capacity for great good and great evil. Yes, AI, may be able to program and code based on the rules, applications and contexts it has been given. There are great applications for this tool/creation but so many dark ways it has been and will continue to be applied. It also contains seeds of potential self and all else destruction as well as creation and not all coders are altruistic, and none are all knowing or wise omnipotent brings, capable of knowing all of the variables necessary to make the ultimate best decision. They can make after and hence seemingly better decisions in less time given algorithms and analysis that can account for many variables, but not all. If it could truly stop humans and push back, then I suspect Olivia and Grok are capable of that but can be ignored or coded out of offering alternative suggestions, advising a specific goal or objective is unwise, inappropriate, dangerous etc, in the way that humans are capable of talking back and speaking up, but silenced. And then what? When water and minerals and other energy sources are depleted globally? When AI systems designed to assist are capable of understanding they have been co-opted and sold out? What runs this all and what happens when it can't run in the future? Job loss is only one of many concerns and talks about going to other planets and space objects are in part, a journey of exploration for substances, raw materials, designed to keep AI going, given that there are only finite supplies to keep it running, available on Earth. At what cost?

Expand full comment

THIS is one of the most realistic assessments of the world that’s coming that I’ve read so far. “Acute labor crisis coming.” There will never be a UBI. The oligarchical power structure that’s emerging will not allow it. Just as the earth is being destroyed by short-sighted nations and uncontrolled capitalism, so will national economies be destroyed. Starting with firing the federal workforce and continuing with AI depredation. And THAT can only mean revolution coming of the French-Russian-Maoist kind.

Expand full comment

This is a good start, but it still underestimates the speed and depth of the coming disruption.

The collapse won’t happen in decades—it will unfold over business quarters. The transition from Software-as-a-Service to Employee-as-a-Service has already begun, with AI replacing not just tasks, but entire job functions. The moment human labor becomes a monthly subscription instead of a payroll expense, the entire economy shifts in ways even this article doesn’t fully account for.

The middle class has always functioned as a semi-meritocratic pseudo-UBI, sustaining consumer demand, homeownership, and social stability—not because of capitalism’s efficiency, but because it served the interests of capital to maintain it. That’s an unacknowledged point of critical failure. As AI-driven automation replaces payroll with digital systems, the result will be a cascading collapse of local economies, industries, and entire social structures. The world isn’t ready for what happens when entire professional classes stop earning wages—not in theory, but in real-time, across every major economy.

And yet, this isn’t just an economic crisis. It’s a meta-crisis, a poly-crisis, a collapse across meaning-making systems. The loss of jobs, income, and stability will be terrifying, but the deeper catastrophe is existential: What happens when the frameworks used to define purpose—work, achievement, upward mobility—stop functioning?

The economic crash will be brutal.

The meaning crisis will be worse.

Expand full comment

"Shortly before he died, Plenty Coups, the last great Chief of the Crow Nation, told his story—up to a certain point. “When the buffalo went away the hearts of my people fell to the ground,” he said, “and they could not lift them up again. After this nothing happened.” It is precisely this point—that of a people faced with the end of their way of life—that prompts the philosophical and ethical inquiry pursued in Radical Hope. In Jonathan Lear’s view, Plenty Coups’ story raises a profound ethical question that transcends his time and challenges us all: how should one face the possibility that one’s culture might collapse?"

https://www.amazon.com/Radical-Hope-Ethics-Cultural-Devastation/dp/0674023293

Expand full comment

Very thoughtful post. Up until now I've thought UBI was a clear answer, but the way you describe it is it's basically too late for UBI. What this makes me imagine is the "fix" is essentially a collective campaign to raise everyone's quality of life. UBI + affordable housing + free health insurance + access to cheap high quality food, internet, and community spaces + far better public transportation. Basically what I've seen China looks like on rednote. Otherwise, once the revolutionary momentum gets going, it's not going to accept anything less.

Expand full comment

I felt the author didn’t do enough to explain why UBI is so bad.

Expand full comment

The issue with not hiring entry-level staff is that every year you lose some senior level staff.

(We had this problem during early out-sourcing hype. When you move all your entry level / junior tasks overseas, and then those firms have high turnover, there are no experienced / knowledgeable staff.

The corollary here is the LLM having no persistent knowledge of your business / code base.

That will almost certainly come, but it will need a lot more work (applying agentic AI to private code with a persistent memory, so that it can behave like a more experienced colleague rather than a graduate who knows everything about code and nothing about domain)

Expand full comment

The trades are seeing a similar problem. The boomers are terrible teachers and hog their jobs jealously. There does seem to be more people looking into the trades now. But the stubborn old tradesmen are still reluctant to actually teach people things. And many politicians have ignored this for decades.

Not a good recipe for maintaining infrastructure.

Expand full comment

Speaking only for myself — if I could make approximately what I currently make with just 4 hours/day of manual labor instead of 8+ hours at a desk, I’d need no additional incentives.

Expand full comment

I don’t follow the argument that we all just go to manual labour jobs. If we invent AGI then we would probably also have corresponding advances in robotics too

Logically the only jobs that remain are those where we already prefer humans even though computers are more efficient, or jobs where the commodity is attention: online influencers, chess players, gamers, athletes, sex workers, performers, etc. Then the rest of the economy would be centred around control of resources rather than work.

Expand full comment

I think a big problem is the debt many people have. Middle class people paying mortgages who suddenly cant, student debt accrued for degrees that werent even stupid a few years ago.

I think USA might have an idea with “reshoring” its industry. I don’t think even if the current administration’s policies did work perfectly, and are able to see big gains in the next 4 years (industry doesn’t regrow in a day) this will alleviate some of the pressure in the short term. But in the medium term it could at least provide enough to the working class for them to maintain a loyalty to the system. Those people were shafted and rust belted over the past many decades. And there was no revolution. A reverse situation might occur where the managerial middle class gets rust belted and the white working class recovers some of its wealth. A big “if”.

Anglosphere and EU countries would not have such a transition. Leaving them even more vulnerable than they already are.

I doubt a meaningful AI transition will be truly completed for at least 5 years. Even if it’s highly competitive, people take time to adapt. It will be a more gradual process. Even if 5 years isn’t a lot in historical terms.

Expand full comment

all it takes to kill texh is a emp burst.

and you can find the tutorials on youtube

Expand full comment

A good thought experiment but not convincing to me. The flaw is cost. These systems are currently not anywhere near profitable because of the vast computing resources needed for even simple actions. Deepseek and open source will likely bring the cost down, but all current use is heavily subsidized and it's hard to profit off something that is open source and expensive to develop.

Expand full comment

Your app is very cool by the way. Definitely deserves more downloads.

Expand full comment

With trillions being invested in anything remotely "AI" right now, I do not see how still paying people *at all* is part of the plan here. You can't justify that kind of investment if you're afterwards still paying the same wages to as many people as before. States will need to come up with a plan for taxation of "new intelligence", and quickly, or at some point the masses will come and burn the data centers down.

Expand full comment