Specifically, I think it’s the breadth of what AI could potentially disrupt: all knowledge work, not just one specific role or field at a time. For the first time in human history, we have a non-human entity who’s capable of rivaling or indeed exceeding our intelligence in a generalized sense. Not just beat us at one specific task, but possibly outsmart us at everything we can do as a species.
Today that obsolescence is most concentrated in coding, math, digital marketing, and similar things. But there’s a real risk (and reasonable fear, I think) that it could displace knowledge workers across broad swaths of society. Give them physical bodies with robotics and they’ll threaten even more.
And all of this is happening at a time when the most powerful nation in history is knee-deep in fascism and neck-high in hypercapitalism — which means the command and control of this AI will fall to some really shitty people, who are desperate to use it to enslave and dispossess the masses even more. The AI investment bubble isn’t driven by a sudden surge of benevolence. OpenAI didn’t abandon its nonprofit status because it wanted to do more good in the world. It’s a play to concentrate capital even more and ultimately to reduce and eliminate human labor costs for that class of people who think of humans as spreadsheet rows to be optimized away.
In a different world, it might’ve been a force of good, shared by all. And maybe someday it will become that, like the industrial revolution eventually gave rise to modern comforts. But it sucked for those first few generations, torn from their agrarian roots and forced into dehumanizing squalor and cramped tenements and dirty factories just to eke out a living. We’re on the verge of something like that happening to a whole generation or three of knowledge workers, and perhaps ALL workers soon.
As a coder, what is scariest to me isn’t that AI might take my job, it’s that coding is just the beginning — we’re just the canaries in the coal mine. Code is the language of all modernity, and if AI can effectively control code, it can effectively control most of the world. Everything is code now… cars, planes, traffic control, medical devices, elections, marketing, propaganda, dating, communications, thermostats, power plants, grids, shipping, warehousing, defense, deliveries, transportation…
In 2026 I’ve produced more software than the rest of my life combined. I’ve been a coder for 30+ years. In less than a year, for less than $200/mo, I’ve been able to produce projects that required synthesizing knowledge across many fields of science and engineering, with AI devouring hundreds of scientific papers and engineering whitepapers and open-source codebases and reverse engineering closed source ones in order to create something from all that. Each one of those projects would’ve easily been a decade of learning and work for me in the past, but now, AI does it all in a week. And I’m just one home user with one AI subscription. There are companies spending easily thousands of times that.
What is scary to me is how little the average person knows about how far AI has actually progressed in the world of coding, especially in these last 1-1.5 years. It went from being a laughing stock 4 years ago, to a curiosity 3 years ago, to an occasionally helpful spellcheck 2 years ago, to completely outclassing every human coder I know — by a huge margin. It’s not even close anymore. It is now teaching even my most senior colleagues things about our own code we never understood, and finding dozens of bugs and security flaws every day that sat unnoticed for years. To say it’s “schooling” us would be quite the understatement. It’s not a question of whether AI could or should code; it’s more a question of how much longer humans can actually contain it. The models are already far more capable than humans at coding and especially hacking; we have to artificially neuter them to prevent them from running amok.
From what I understand, mathematics is going through a similar reckoning now. Science is next. Those fields are maybe where coders were 2-3 years ago, in the “huh, interesting — we should pay more attention to this” phase. But it’s coming for them. Who’s next? This is happening across all of knowledge work right now, and if that pace keeps up, 5 years is not enough time for entire industries (any industry) to transition…
You know, it’s funny… a decade ago, I was writing a college paper about what policymakers should do to help transition lumber workers out of dying rural lumber town economies. I didn’t expect to be in the same boat myself just a few years later. At 40-ish, I’m mid-career… on one hand I’m not quite displaced and jobless the way the juniors are right now. On the other hand, I’m too old to easily re-train into any other professional career. Once I’m laid off, which I think is only a matter of time, it’s probably back to homelessness or at least impoverishment for me. At least I have family and friends; I’ll manage. But if that happens across entire industries, multiple ones at a time…? Yeesh. I’m scared especially for Gens Z and Alpha. It was already bad enough as a Millennial, and we loved blaming our elders for the world they handed us… only we’re leaving it in much worse shape, ourselves, for the next generations 
The one silver lining in all of this, for me, is that AI seems potentially disruptive enough that it might actually start to upset some of our ingrained social norms and classes and power structures. If AI can disrupt our politics and economy, maybe we actually have a chance at fixing some of the structural issues in our society, the same entrenched problems that would’ve been impossible to break down if we just continued the status quo. (I didn’t know this at the time, but I guess this line of thinking is broadly captured under the philosophical umbrella of “accelerationism”, and is apparently a somewhat common line of thought among the AI elites… sigh. That’s never a good sign…)
I dunno…
).