This weekly roundup thread is intended for all culture war posts. 'Culture war' is vaguely defined, but it basically means controversial issues that fall along set tribal lines. Arguments over culture war issues generate a lot of heat and little light, and few deeply entrenched people ever change their minds. This thread is for voicing opinions and analyzing the state of the discussion while trying to optimize for light over heat.
Optimistically, we think that engaging with people you disagree with is worth your time, and so is being nice! Pessimistically, there are many dynamics that can lead discussions on Culture War topics to become unproductive. There's a human tendency to divide along tribal lines, praising your ingroup and vilifying your outgroup - and if you think you find it easy to criticize your ingroup, then it may be that your outgroup is not who you think it is. Extremists with opposing positions can feed off each other, highlighting each other's worst points to justify their own angry rhetoric, which becomes in turn a new example of bad behavior for the other side to highlight.
We would like to avoid these negative dynamics. Accordingly, we ask that you do not use this thread for waging the Culture War. Examples of waging the Culture War:
-
Shaming.
-
Attempting to 'build consensus' or enforce ideological conformity.
-
Making sweeping generalizations to vilify a group you dislike.
-
Recruiting for a cause.
-
Posting links that could be summarized as 'Boo outgroup!' Basically, if your content is 'Can you believe what Those People did this week?' then you should either refrain from posting, or do some very patient work to contextualize and/or steel-man the relevant viewpoint.
In general, you should argue to understand, not to win. This thread is not territory to be claimed by one group or another; indeed, the aim is to have many different viewpoints represented here. Thus, we also ask that you follow some guidelines:
-
Speak plainly. Avoid sarcasm and mockery. When disagreeing with someone, state your objections explicitly.
-
Be as precise and charitable as you can. Don't paraphrase unflatteringly.
-
Don't imply that someone said something they did not say, even if you think it follows from what they said.
-
Write like everyone is reading and you want them to be included in the discussion.
On an ad hoc basis, the mods will try to compile a list of the best posts/comments from the previous week, posted in Quality Contribution threads and archived at /r/TheThread. You may nominate a comment for this list by clicking on 'report' at the bottom of the post and typing 'Actually a quality contribution' as the report reason.
Jump in the discussion.
No email address required.
Notes -
Trillions of dollars are being spent on building datacenters for inference. Amazon software engineers are inventing bullshit work for AI to inflate their internal usage scores.
I’m no expert, but isn’t there a fatal flaw here? Most of the work LLM inference is used for is essentially busywork that wouldn’t exist in an automated economy. It’s writing emails, it’s code reviews, it’s asking dumb questions, it’s transcribing or summarizing research or zoom meetings. Even in software engineering, a lot of LLM tokens are used in the kind of inference that a hypercompetent solo-coding model with limited or no human oversight just wouldn’t need.
Think of an office with 10 human employees working in, say, payroll, constantly sending each other emails, messages, having meetings, calling and speaking to each other and other people, summarizing documents, liaising with other departments, asking AI question about how to use various accounting tools, or about the company’s employee benefits package. Now say this department is automated. An AI model acts as an agent to use an already-existing software package to do all the payroll work. No emails, calls or meetings - or at least far fewer. The total inference work required goes down. And the existing software package doesn’t use AI (even if it may have been coded with it), because you don’t need AI to compute payroll data once you have sufficiently complex and customized software for your business.
In the same way, if we imagine our automated future, super high intensity / high token usage inference is actually not really universally required in a lot of occupations. It will be for some multimodal work (plumbing, surgery, domestic cleaning in complex physical environments), but for many tasks, one-and-done software coded either by AI or that already exists can just be deployed at low intensity by an agent. The AI that replaces your job might at first do a lot of coding, but as time goes on, the amount of novel inference required will diminish. Eventually, software coded in a one-and-done way by the AI may actually handle almost all the workload, and token usage for generation may be very limited to just some high level agent occasionally relaying instructions or performing oversight.
In this scenario, why would we expect inference workloads to shoot up so dramatically? Much enterprise AI usage is currently “fake” in the sense that it would not be performed in a fully automated environment. It’s a between-times thing.
It is surprising how much can you achieved with good prompt and harnesses nowadays with how little tokens. The problem is that the majority of people using AI are too stupid to be lazy in the proper ways. I think that a tornado is coming. Probably later than anticipated, but the white collars brains are afraid (insert starship troopers movie meme here) - especially the ones who deep down always knew that their intellectual labor is neither extremely intellectual nor much useful. I am already seeing proposals for excise tax on tokens. And I think that the big hyperscalers grossly underestimate how much optimizations are left in the pipeline.
The compute cost on tools is low, agents are becoming quite adept at tool calling - so agents creating their own tools and tool calls is totally expected ... in a way this is what programmers have always done.
There is lots of performance left to be squeezed out of each token. And relatively small hyper focused models also doesn't seem to be getting the attention it deserves.
I'm always amazed at how often this refrain comes up, with different explanations every time. For some reason, he idea of bullshit jobs is one has immense staying power.
Whenever it does come up, I often wonder how one would separate the useless, lazy, stupid jobs from the essential ones. When I was younger I held a similar view, but over time I realized that the single strongest predictor for whether I thought a job was bullshit or not was how little I knew about its actual day to day work.
As a simple example, take project managers. A bad one is terrible, and is probably one of those things that a lot of people woud say is neither "intellectual" nor "useful". I had that opinion once upon a time. Eventually, I worked on a project with a good project manager and realized that they actually do an insane amount of work and provide a significant force multiplier for the rest of the people involved. It felt fantastic to just... work on the problem.
That's one of my biggest concerns about the current LLM frenzy. It's largely being driven by a small, cloistered group of people who really buy into the "bullshit jobs" premise, and spend more time saying "well couldn't you Just X" instead of figuring out why things are the way they are. Systems evolve into specific shapes for a reason. Tribal knowledge is real.
I feel like we're going to be forcefully reminded of those facts if we keep it up.
"Bullshit jobs" is, as far as I can see, one half large organizations being too slow to adjust course when jobs need to change, and one half wishful thinking by utopians who desperately want wage labor to be bullshit so they can make the case for some form of luxury communism.
It’s a useful way of describing work that has been regulated into existence. For example, the EU passes legislation that requires some hugely complex and time consuming climate reporting for every company with an annual revenue of more than €10m. 100,000 companies now have to hire someone to be their ‘climate reporting officer’. The US healthcare system’s extensive regulation and lifetimes of case law about who pays and when and what insurance covers and what the hospitals have to provide etc etc create tens of thousands of jobs on both sides of the billing equation (the healthcare providers and the insurers) that don’t exist, or certainly don’t exist in the same sense, in single payer systems. Walmart wants to open in a town in Kentucky. The town offers large tax breaks in exchange for hiring 200 local people. A big Walmart in 2026 only needs 120 people to operate, though, but the tax breaks are worth more than that payroll. Numerous jobs as greeters and shelf stickers and security guards are created unnecessarily. A government contractor is tasked by a new government with proving that what it does at $500m a year in state billing is justified. It hires McKinsey for $20m to write a report, because nobody ever got fired for hiring McKinsey (including the minister who gets the report).
Individually these are examples of bloat, bureaucracy, overregulation, unintended consequences, inefficiency, corruption, graft, credentialism, whatever. But collectively, all of these are examples of bullshit jobs.
If the bureaucracy is being imposed from within the corporation, it's one thing, but it's totally different if it's a necessary response to legislation. At that point it's less about the job itself being bullshit and more about disagreement with the underlying policy. If the job performs the function of complying with the law, it's a fairly large value add compared with the penalties that would be imposed if the work weren't done. To give an example of a regulation that can come across as bullshit to some people, the EPA requires erosion and sedimentation (E&S) permits for construction projects that involve disturbing a certain amount of earth. Depending on the size and location of the project, you may need to apply for a permit, not need anything, or need to have an E&S plan on site but not need prior approval. This third category can come across as bullshit to some people, because it involves paying an engineer thousands of dollars to publish a report that no one is going to read, especially if the conclusion is that no special precautions involving erosion need to be taken.
You could just as soon not get a plan and no one would be the wiser. Except if runoff from the jobsite ends up washing onto your neighbor's property and he asks to see the plan and you don't have one. If you end up getting sued over excessive runoff causing damage, not having a plan to deal with erosion is a pretty big matzo ball to have hanging over the litigation. Sure, the government could eliminate E&S requirements entirely, but that only means that when a problem happens you get to spend several years litigating it. The tradeoff is that you minimize erosion problems on all projects from the beginning, and if you do get sued it's nice to be able to say that you had an E&S plan.
The problem I have with the bullshit jobs theory in general is that somebody who isn't familiar with a business presumes that they know how to run it better and knows what work contributes value and what doesn't. This is the fundamental issue I have with AI gurus saying that LLMs are going to take your job. Really? Because chances are they have no idea what you actually do, let alone what value it provides the company. They think of everything in terms of outputs and assume that being able to generate the output is the beginning and end of the value the employee provides to the company. It's a prime example of Rory Sutherland's Doorman Fallacy: A consultant to a hotel company sees the doorman's job as opening the door, and he tells the hotel that they can save a ton of money by replacing the doorman with an automatic system. But the doorman does more than open the door. He calls cabs, he deals with package deliveries, he provides a certain amount of security, he gives the hotel a degree of prestige, etc. Since it's impossible to quantify how much business you're getting as a result of these little services, it's easy to fall into the trap where you believe that automating away the doorman is an automatic windfall, especially when nobody is ever going to say in a customer survey that the existence of a doorman played any role in selecting the hotel.
I’m more amenable to the idea that some jobs are bullshit. It happens mostly by inertia— we’ve always done it this way, we’ve always had a person to do X thing, so we still need that person doing that thing. Yes you can have value added — people doing a service oriented thing often make the experience of purchasing something a bit nicer. A food-o-mat existed in the 1950s, you simply punk in money and the food would be put behind a little door and it all worked sort of like a giant vending machine. Heck we still have actual vending machines, and you could easily create a food selling business that worked almost entirely by stocking vending machines. But you don’t lose the waitress because there’s simply something pleasant about buying something from a person who makes the experience pleasant. That would require at least some premium to the service. A consumer would have to want to pay more for a person to do that. And for customer facing roles, sure. But the same cannot be said for backend types of work. There’s no reason to pay extra to have a secretary type up your messages and emails. There’s no benefit to having a human make a spreadsheet. No one cares whether their balance sheet was created by a human. So those jobs are more at risk because they don’t get any better because the job was done by a human who made the experience nicer.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link