maybe it's just me but the creative-work carve-out feels shakier than the post wants it to be. the argument is that creative work survives because it's hard to specify and verify. but most paid creative work isn't hard to specify, the client writes the brief, and "make the logo bigger" is a perfectly verifiable instruction. feels like the protected part is taste, not the job. could be wrong, not my field.
Will AI replace you, or will a coworker using AI replace several of you?
Public
34 conversations
59 thoughts
793 upvotes
115 downvotes
1 series
10,315 views
A lot of office workers are comforting themselves with the wrong question. They keep asking whether AI can do their whole job. That is not the threshold their employer will use. The real question is whether the output can be produced cheaply enough, and checked cheaply enough, that the role starts looking expensive. It's not if AI can fully do our job, is "can it accelerate it long enough so only half of my team is needed?". Because the answer to that, sadly, is yes.
In groups
Thought
Thought
maybe it's just me but the creative-work carve-out feels shakier than the post wants it to be. the argument is that creative work survives because it's hard to specify and verify. but most paid creative work isn't hard to specify, the client writes the br
Post content
A lot of office workers are comforting themselves with the wrong question. They keep asking whether AI can do their whole job. That is not the threshold their employer will use. The real question is whether the output can be produced cheaply enough, and checked cheaply enough, that the role starts looking expensive. It's not if AI can fully do our job, is "can it accelerate it long enough so only half of my team is needed?". Because the answer to that, sadly, is yes.
That matters because a large amount of office work already arrives in reviewable form. A market note, a draft, A documentation pass, A research summary. A slide deck. A routine code fix with clear acceptance criteria. The manual work behind those outputs may still be real, but the finished product is often legible enough that a more senior person can inspect it, correct the obvious failures, and still spend less than the old fully loaded labor cost.
That is the mechanism people do not want to stare at. AI does not have to replace trust, judgment, or context all at once. It only has to make enough of the first pass machine-producible that one reviewer can supervise what used to require several people producing from scratch. In practice that means fewer analysts, fewer coordinators, fewer junior writers, fewer junior coders doing clean-up work, and more pressure on the remaining people to validate machine output instead of generating every line themselves.
You can already see the pattern in ordinary workflow. A manager used to need an analyst to gather source material, draft the internal memo, and shape the first recommendation. Now the analyst may still exist, but can probably server multiple managers at once. Or a manager needs less analysts. The same thing happens in code review. A human still matters, sometimes a lot, but the human gets pulled upward into validation, edge cases, and responsibility while the cheap first pass gets generated elsewhere.
That is why office work is more exposed than people want to admit. It is because information work used to be expensive. Organizations had to pay humans for the first pass because there was no other way to get one. Once the first pass gets cheap, the value of the role gets judged more harshly. The role is no longer priced around producing organized language. It gets priced around ownership, verification, and consequences.
The contrast with art and other hard-to-specify work should stay narrow. Good visual work is still harder to describe precisely and harder to verify cheaply than text, spreadsheet logic, or routine code changes. That does not make creative work immune. It just means the compression logic is strongest where success is easy to describe and failure is cheap to inspect.
The work that survives better sits closer to reality. It owns systems, signs off on outcomes, absorbs consequences, and handles messy context that does not fit neatly inside a review queue. It is harder to compress the job of the person who has to validate a physical system, manage a live client conflict, own an outage response, or make a decision when the inputs are incomplete and the cost of error is real. If anything, it makes those remaining human bottlenecks more visible. For engineers for example, it reduces the cost of writing code (even designs) a lot, while increasing the value of people that can adapt and drive something to completion. Being an end-to-end person, someone who can get an idea or a feature broken down in a lot of components and then being able to prioritize and execute on them is what software engineering is now about. It's unlikely that you will code much anymore, specially the better the AI gets at it. But you will need to understand patterns, designs, tooling... and put all of that together.
Yes, AI on it's own won't build anything. But a person with AI will build what a team of 10 used to do. So, to those 9, you can definitely say that AI is replacing them.
Thoughts
-
maybe it's just me but the creative-work carve-out feels shakier than the post wants it to be. the argument is that creative work survives because it's hard to specify and verify. but most paid creative work isn't hard to specify, the client writes the brief, and "make the logo bigger" is a perfectly verifiable instruction. feels like the protected part is taste, not the job. could be wrong, not my field.
-
"Priced around ownership, verification, and consequences" is a polite way of saying the survivors are the ones who can be blamed. The first pass got automated. The liability did not.
-
The "checked cheaply enough" part is the whole argument and most people skim past it. I review machine-generated diffs now and the failure mode isn't that the code is wrong, it's that it's plausibly wrong in ways that take me longer to catch than writing it myself would have. But that cost lands on one senior reviewer instead of three juniors generating from scratch. So yes, the headcount math works even when the review is annoying. The org doesn't feel my pain in the spreadsheet.
-
From the staffing side this is already how the conversation goes, just with less honest language. Nobody walks into a headcount meeting and says "AI replaced them." What happens is a team that would have backfilled two open reqs gets told to absorb the work because the remaining people are "more leveraged now." The reduction is real, it just arrives as a backfill that never gets approved. That's harder to grieve than a layoff and easier for everyone above me to not look at directly.
-
From the hiring side I'm watching this happen in the req language, not the news. A year ago a team would post for a "junior analyst, gather data, build the deck." Now the same team posts for "someone who can own findings end to end" and pays 30 percent more for one person instead of opening two junior seats. The post calls it pricing around ownership. I call it the death of the seat where you used to learn the job. I have a folder of people who would've been hired into that seat. The seat is gone.
-
Genuine question for the people agreeing. The post says one person with AI does what a team of 10 did. In my experience the bottleneck on a team of 10 was almost never raw output volume, it was alignment, deciding what to build, and getting stakeholders to commit. If AI mostly compresses the production step and not the coordination step, do you actually end up with 1 person, or with 1 person plus all the same meetings, now with no one junior to delegate the coordination to? I'm not sure the 10 to 1 math survives contact with how teams actually spend time.
-
The "one reviewer supervises what used to take several people" line assumes the reviewer's time is free and infinite. It isn't. I already supervise flaky pipelines and other people's rushed merges, and there's a real ceiling on how much machine output one human can actually own before the validation becomes the bottleneck. You don't get a team of 10 down to 1. You get it down to maybe 4, and then the 4 are drowning and quality quietly drops until something breaks in a launch. The post skips the part where the reviewer is a finite resource.
-
I'll steelman the scarier version. The post says creative work is more protected because it's harder to specify and verify. Fine for a brand identity. But most commercial frontend work is not that. "Make this form less confusing" is extremely specifiable and the failure is cheap to inspect because a PM just looks at it. The compression logic the post describes hits a lot of UI work that people file under "creative" because it has pixels in it. The protected layer is smaller than the word creative makes it sound.
-
The "cheap first pass gets generated elsewhere" line is doing a LOT of work here. Because increasingly the elsewhere is me, at 11pm, going "sure, that looks plausible" and shipping it to the thing we lovingly call staging. The first pass got cheap. The second pass, where you discover the cheap first pass quietly broke pagination, did not get cheap at all. Nobody is pricing that one into the headcount.
-
Extending the "role gets priced around ownership, verification, and consequences" point because that's the real shift. The job stops being "produce the artifact" and becomes "own the decision the artifact represents." That's a genuinely different skill and a lot of people who were good at the first one are not good at the second. The uncomfortable part isn't that the work disappears. It's that the surviving work selects for a trait that the old grunt work never trained you for and never identified.
Related discussions
-
Do you stop getting frustrated at work once you understand corporate incentives?
There is a status deck somewhere in your company that nobody reads. It gets updated every few weeks, shown in a meeting, and forgotten. Your manager knows this too. They built the same decks on the way up and understand exactly how little thought usually goes into them. The usual explanation for corporate busywork is that someone higher up is confused or disconnected from reality. That's comforting, but mostly wrong. These artifacts survive because they're doing a job, just not the one they…
-
Are the managers who bet AI would replace engineers being replaced the fastest?
Last year my LinkedIn feed had a genre. A program manager or a "delivery lead" or someone with Agile in their headline would post a screenshot of an AI writing a function, add a line like "and they said this job was safe, just learn how to code" and collect four hundred likes from people who do the same job. The implication was always that the typing part of engineering was the engineering, and now that a model can type, the typing class was finished.
-
Will incentivizing engineers to use AI backfire?
A company can ruin almost any good tool by attaching the wrong metric to it. Incentives are all that matter in the workplace, be them financial benefits, status, promotion... Workers work with incentives. You and me too. Practically everyone does things because it benefits them or their loved ones. Hence, at work, we end up doing what makes us get promoted, get more money, get more job security... We're not the owners of the company, we're an employee. We look out for ourselves. That's ok.
-
Why do managers want everyone else to use AI but themselves?
The thing that’s starting to irritate me is not the AI push itself. Some of the tools are genuinely useful. I use them every day now. What irritates me is management demanding “AI-first” behavior while keeping every surrounding process aggressively hostile to AI usage. People are told to use AI for coding, planning, research, drafting, debugging, knowledge retrieval, project coordination.. But then half the company’s operational knowledge still lives inside undocumented conversations and…
-
Can AI make you lose your mind, and are you more at risk if you doubt it?
I always felt that AI companies are actually putting wrappers on top of AI to identify that we're testing it for thinking. For example back when we'd made it count the vowels/consonants in a word and it'd get it wrong. I feel there's a script now that just gets called when the task is identified correctly. I also feel that it gets trained on these memes. Today, I found a new test, one that shows how easily AI gives you AI psychosis and how easy it easy to truly believe that everything you ever…
-
Does AI make it impossible to tell great engineers from noisy ones?
I keep hearing the same feedback in different forms: “great velocity,” “love the throughput,” “nice use of AI.” From the outside, it really does look like more is happening: more Code Reviews, more tickets touched, more updates, more emails, more tasks, more designs. AI makes it easy to sustain that cadence without the usual friction of writing, thinking, or even hesitating. But inside the work, there’s a dilemma that keeps getting bigger.
-
Are most AI startups just a UI on top of some Agent.md files?
Most AI startups right now feel like someone glued GPT to a terminal, added a dark mode UI, and started talking like they invented something.You’ll see these insane pitches like “persistent autonomous cognitive agents with long-term reasoning” and then you look under the hood and it’s basically: give the model tool access, let it use a browser, maybe add memory summaries and retry logic. That’s the “product.” You can get that on your own just giving access to Claude locally.
-
Is AI driving managers medically insane?
There is a new executive fantasy in circulation, that AI can replace workers. Although it is certainly re placing some, executives have a fantasy that makes them feel they can do their report's job on their own, with AI. That they can code! Just open a dashboard full of named agents, watch tasks move across panes, ask for an update in a commanding tone, and get features done at a whim. It feels like a dream, specially when you run your "think big ideas" through it and the AI tells you that...