The AI jobs-apocalypse is here: Prepare for the laptop-class revolt
Zickgraf (Ryan)
Source: UnHerd, 17 Feb 2026; partly reprinted in The Week Issue 1582. 7 March 2026
Paper - Abstract

Paper StatisticsBooks / Papers Citing this PaperNotes Citing this PaperText Colour-ConventionsDisclaimer


The AI jobs-apocalypse is here Prepare for the laptop-class revolt1


AI in the UK: how will jobs be affected?34

36 UnHerd Comments51
  1. George K: 23 days ago
    Volatility yes but it smells bubble in the air. It’s been a year without any significant progress as LLMs apparently exhausted their potential. I use AI daily for my job and it’s an awesome technology which gives me a boost of about 20% productivity but it’s nowhere close to telling me that I better travel around the world rather than troubleshooting deployments
  2. Andrew Vanbarner: 22 days ago
    Reply to George K.
    Same here. AI will enhance productivity – I agree with the estimate of about 20% or so – but it still requires the user to ask precise questions, and interpret the answers correctly. The “what” of data issues is readily apparent – six isn’t nine, for example. The “why” – why is the output negative six million, when it should be nine million – isn’t as easy to deduce, mitigate, or solve. People who spend their days calling out the “whats” could’ve been automated some years ago. People who can get the “whys” built, fixed, or functioning are unlikely to ever be replaced.
  3. Matt Sylvestre: 21 days ago
    Reply to George K. LLMs have exhausted themselves yes but now the real punch is in application which is, for now, lagging behind… Also, there will be another breakthrough that moves the AI ball forward, its more a matter of when than what…
  4. Louis Biafore: 23 days ago
    As someone who works one the technical side of building agentic systems that perform work previously done by humans I found this article both very ‘on point’ and pleasantly original in its synthesis. Right up until the last 3 paragraphs (“As AI automates…”), at which point the exposition careens wildly outside the bounds of what seems technically plausible.
  5. UnHerd Reader: 23 days ago
    Good article, but needs more precision. Take the USA, or UK for that matter. Take New York, or London for that matter. The data show that the jobs most affected will be female-dominated and entry- to mid-level — finance, insurance, HR, legal services. In truth, AI can do a better, faster job already. The data also show that these sectors are populated by already rather radical people (insert great feminisation thesis) who are majority “progressives.” So what does the future hold? A shrinking population of childless, 40 plus, urban dwellers with no real skills beyond mastery of memos. What we may get is a horrible transition period where Rachel from accounts re-manifests herself/they-self as a “change agent.”
  6. Lancashire Lad: 23 days ago
    Reply to UnHerd Reader: Quite possibly; the really interesting points made in this essay (RK’s best for Unherd imo) revolve around the entire reshaping of our political landscapes as a result of job losses and the posing of perhaps the greatest existential question of all: what are humans for? Throw into the.mix that the political views of Chinese and Russian citizens are already ignored and suppressed, giving their economies a head start once the repercussions start to unfold. So also: how do we retain our relative freedoms, in a new age when ‘work’ is no longer a means of social agency?
  7. UnHerd Reader: 23 days ago
    Reply to Lancashire Lad. Excellent question. I’m sure we have different answers. For me, this cuts to the contemporary problem of the institution and corporation as sites of both sociality and creativity. The traditional bureaucracy and the large corporation may become “victims” of this historical moment, but the university is the space to watch. I think we’re already seeing the changes, and the collapse. What comes next will be, minimally, interesting to watch.
  8. Lancashire Lad: 23 days ago
    Reply to UnHerd Reader. Could be that actual “answers” aren’t something that can be currently addressed. The one thing that we have as humans is our very long history of adaptability. I’m sure if we can set out across the ocean in a flimsy craft, not knowing whether there’s anything else out there; or if we can head further and further north in search of food until we learn to survive amid snow and ice rather than baking sun, we can learn to adapt to the changes that we’re bringing upon ourselves. Our ancestors didn’t have “answers” to hand when they did those things, but we survived, we thrived and I don’t think we’ve actually changed very much in our cores. This is what i refer to when I speak of ‘the human spirit’, stretching back many millennia.
  9. Charles Stanhope 22 days ago
    Reply to Lancashire Lad. Trust that Porton Down is working on a ‘new’ “vaccine”, and one that bl**dy well works this time, unlike the Chinese rubbish. We face a ‘Brave New World’ where only Alphas need survive, Betas and Gammas will be totally superfluous. The environmental and social benefits will be simply stupendous.
  10. John Ellis 22 days ago
    Reply to Charles Stanhope. But like Huxley’s original, it will soon metamorphise into a dystopia. And even he had space for Betas, Gammas and Epsilons. Although your “vaccine” will need to be advanced indeed if it is to cull humans on the basis of their bank balances….
  11. Charles Stanhope: 22 days ago
    Reply to John Ellis. Being a patriotic Englishman I was thinking of something along the lines of ethnicity. It seems both the simplest and eminently beneficial solution. Huxley was writing in the 30’s as you well know, so was at a considerable disadvantage as far the advances in science are concerned. However had he been alive today I am certain he would have proposed something very similar. “It cruel to be kind” as we used to say.
  12. UnHerd Reader: 23 days ago
    Reply to UnHerd Reader. Feminization Reset. As per the above noted commentator, under this scenario AI doesn’t bode well for the class of “progressives”. Perhaps this drives the elite WEF globalists to reduce the world population from 8 to 2 billion. AI, the epitome in digitization, as the 4th Industrial Revolution. Isn’t this a contradiction, an oxymoron, in terms!
  13. Matt M 23 days ago
    Reply to UnHerd Reader. The answer to that problem is marriage. As women’s white collar work disappears, they always have the option to marry a builder, live off his (now elevated) salary and have a few babies. Which is what she probably wanted in the first place (which is why she/they is so prickly and unpleasant at the moment).
  14. steve hay 21 days ago
    Reply to Matt M. We are already seeing the prospects for tradesman and women to be looking on the up and up. You will already pay through the nose, comparatively speaking for the services of a competent “Tradie” and good luck to them. How interested they would be in a snarky disagreeable partner is another thing. They tend to be practical people and an entitled, probably unemployed potential partner, is highly likely to be given a miss
  15. Catherine Johnson 23 days ago
    Reply to UnHerd Reader. You do wonder if more women will simply choose to be stay at home Moms in this new environment – particularly as more and more discussion is already taking place about how empty the modern feminist boss babe vision has been for many women.
  16. Louis Biafore 23 days ago
    The author has seemingly lost track of the critical distinction between jobs the current LLM-based agentic systems can do: manipulation of symbols in constrained environments (paralegal work, programming) and those they cannot do, namely those which involve manipulating the physical world in unpredictable environments (plumbing, ambulance paramedic). There simply have been no breakthroughs in robotics comparable in magnitude to the transformer architecture, introduced by Vaswani et al in their seminal 2017 paper, “Attention is All You Need”, which underpins LLMs and the agentic systems being built upon them. Discussion for another day, but it does look like “pink collar” jobs are apt to be more affected in this round than “blue collar” jobs.
  17. RA Znayder 23 days ago
    The second part of this piece explores intriguing possibilities. But for that, the first part has to be true first. And that is not really the case. So far, there really is no AI job-Apocalyps. Reports that look into it show that jobs actually being taken over by AI are quite rare. However, there is no shortage of companies that claim that their downsizing operation is due to AI. It is what investors love to hear. If anything the reasons are economic. The inflationary asset bubbles, of which AI is the biggest, are squeezing real demand. Also never talk to AI CEO’s. Most of what they have been spreading is PR; they have been selling hypothetical fears for years. Apparently investors love that too, otherwise they would not do it. Instead, talk to researchers and engineers actually working on scaling and persistent AI-problems to get a good picture. “On the far Left, we see a rise in techno-scepticism and a demand for the abolition of “bullshit jobs””. This is an interesting one. Because I feel that many of the managerial jobs that AI could potentially take over are already bullshit jobs. So if it is not really about the output of these jobs, there is not much for AI to take over.
  18. George K 23 days ago
    Reply to RA Znayder
    Both good points about companies claiming fashionable AI based reductions and AI CEOs being unreliable source of information. Unfortunately I’d add also all AI “godfathers “ ( there’s about as many as podcasts) I never heard any supporting evidence for extraordinary claims
  19. Louis Biafore 23 days ago
    As someone who works in the “agentic” space it looks to me like many jobs—those that involve operating out in the physical world, especially those that require manual dexterity—will only be affected at the margin by LLM-based agentic systems. Even some of the “bullshit jobs” (the author obliquely references the late David Graeber’s book of the same name) will survive in some form. Of Graeber’s 5-fold typology of bullshit jobs: (1) flunkies, (2) goons, (3) duct tapers, (4) box tickers and (5) taskmaster, only “box tickers” look doomed to me. Certain “flunkies” will almost certainly survive—it will continue to be a signal of prestige to retain a human doorman and human receptionist, though you may need to be a really attractive PhD from Cambridge to get the job The second Graeber category, “goons”, which includes lobbyists will likely survive as long as nation governments have budgets. The third category, “duct tapers”, which includes programmers who fix things that should have worked without their intervention, will survive as long as there are new versions of AI systems, which invariably fail to work perfectly from day one. And Graeber’s final category, “taskmasters”—basically middle-management—will survive in some form unless the oligarch himself plans to spend his day directly dispatching orders to his AI minions. In addition, the entire superstructure of the Western legal and corporate architecture requires at certain points in the organization that there exist identifiable “wringable necks” in the event something goes wrong. As AI’s currently are indifferent to having their (non-existent) necks wrung, these roles will continue to be held by humans. In short, there are plenty of reasons for A to employ B, only one of which has to do with what B produces by their labor.
  20. Andrew Vanbarner 22 days ago
    Reply to Louis Biafore. True. I find AI enormously helpful for my job, but it still doesn’t write particularly clean script for the kind of programming I occasionally do, sometimes just makes things up, and is once in a while flat out wrong while perhaps being technically correct. I’ve yet to see a robot accountant, nor an entirely automated project manager. Nor will AI entirely replace human judgement. AI to me is just a series of math models and algos, which we can now run quickly and efficiently using the cloud and more powerful servers and processing. But insofar as I’ve spent decades with uncooperative coworkers, unreasonable bosses, or unpleasant people who don’t seem to do very much, I won’t be upset about some people’s jobs being automated. The rest of us can lead, follow, or get out of the way. And we’ll still be paid wages if we do so.
  21. Alex Lekas 23 days ago
    “it will move faster through institutions because the people experiencing it — journalists, lawyers, policy staffers — are the ones who produce political legitimacy itself.” Seriously? This self-regard aside, AI is going to produce upheaval and no one should be surprised. The unknown is upheaval to what extent. Too many people treat this as an existential race between East and west. “We have to be #1,” say American pundits and pols, but they’re never clear about #1 in what. It’s as if they’re not sure what the fallout will be, either. The data center mechanics have been interesting. Not many communities want these things in their backyards and I have seen a disturbing trend in which city council members and other public officials sign NDAs about these projects. That borders on the criminal. When the people we elect and appoint are barred from discussing something that will impact the area, that’s a bad sign.
  22. UnHerd Reader 23 days ago
    Anyone believe the public sector will accept AI, when large sections still use fax machines? Link. The result will be that unemployment is centred in the private sector, much of the AI infrastructure will be offshore and the government will continue to expect the same tax take from a shrining private sector. A couple of thoughts … the internet was expected to eliminate high street estate agents. This did not happen? I can guess why – issues of trust in trading assets whose long-term value was seen as rising forever – but I would like to see a proper report. Also, current AI is mean-regressive. The imitation produced by AI is influenced by the full range of available models. Currently, it is easy for someone with genuine expertise to produce better quality work than AI and to use AI to produce the same quantity. This may of course change if AI becomes more critical and less imitative.
  23. Edward Wynn 23 days ago
    This article completely fails to address key issues. The need for human to human interaction which cannot be replaced by IT. The reality that many jobs involve IT as a side or service issue. If you lose your job because it’s been automated by a computer you were wasting your life anyway. The UK for example has a massive shortage of welders and other trades. AI has no role there. If your job involves working on a computer you were always quite likely going to be toast. Middle class bureaucrats especially in the Civil Service will defend their existence to the bitter end. The fact that western country population birth rate is declining is actually a good thing in this context. I’m going to buy a small holding with a good water supply
  24. G M 23 days ago
    Those involved in the trades, plumbing, electrician etc., will be the least affected, until robots can do their jobs. Many of the software, lawyers, office staff tended to vote leftist and they will be affected a lot.
  25. Martin Johnson 18 days ago
    The future is unknown, but I hope that one consequence will be that as demand for its output declines the higher education complex shrinks to less than 1/2 of its current footprint, ideally about 1/4. All the rest has been pure waste, if not destruction of value, for a generation, at least. AI might close the deal if we can free ourselves from the dead hand of the past. Get rid of all the “Studies” departments and programs, all the J-schools, all the graduate-level Ed Schools, half of the B-schools, a quarter to half of the Law schools. And then recast federal funding to not be subsidizing (through overhead rates on research grants, and student grants and loans) most of the humanities and a good portion of social sciences. Some of the savings might be put toward useful endeavors such as engineering and hard sciences. Those who think these moves are bad should pay for those functions with their own money. If states and localities and philanthropists and tuition-paying families want all that, nobody should stop them. But neither should my grandchildren enter life with hundreds of thousands of dollars of debt used to pay for things that will not benefit them. That will be a start on making federal government spending serve rational purposes.



In-Page Footnotes

Footnote 1: Footnote 2: Footnote 3: Footnote 4: Footnote 5: Footnote 6: Footnote 7: Footnote 8: Footnote 9: Footnote 10: Footnote 11: Footnote 12: Footnote 13: Footnote 14: Footnote 15: Footnote 16: Footnote 17: Footnote 18: Footnote 19: Footnote 20: Footnote 21: Footnote 22: Footnote 23: Footnote 24: Footnote 25: Footnote 26: Footnote 27: Footnote 28: Footnote 29: Footnote 30: Footnote 31: Footnote 32: Footnote 33: Footnote 34: Footnote 35: Footnote 36: Footnote 37: Footnote 38: Footnote 39: Footnote 40: Footnote 41: Footnote 42: Footnote 43: Footnote 44: Footnote 45: Footnote 46: Footnote 47: Footnote 48: Footnote 49: Footnote 50: Footnote 51:

Text Colour Conventions (see disclaimer)

  1. Blue: Text by me; © Theo Todman, 2026
  2. Mauve: Text by correspondent(s) or other author(s); © the author(s)



© Theo Todman, June 2007 - July 2026. Please address any comments on this page to theo@theotodman.com. File output:
Website Maintenance Dashboard
Return to Top of this Page Return to Theo Todman's Philosophy Page Return to Theo Todman's Home Page