This article assumes the falsehood that LLMs are actually displacing tech jobs while maintaining productivity. They are not.
What LLMs are doing is being adopted by bazinga capitalists during a bubble, simultaneously providing a tool by which to discipline labor and do layoffs. They are not particularly good tools for actual technical work, one person with an LLM can easily be less productive than that same person not using one. It is telling that companies force LLMs on their technical teams rather than allowing them to be adopted organically as actually useful tools like basically every other tool they’ve ever adopted.
So what this really reveals is that the tech workers, who see their typical work actually producing something tangible (though still to sell some no sense) via their output, see that the entire corporate apparatus doesn’t really care about productivity, they care about chasing profit bubbles and disciplining labor. They will accept the garbage of LLMs and break their product and cut productivity in half while pouring money into Claude, they don’t actually care that it doesn’t work, and therefore didn’t care about your work in the first place.
The grain of truth is that this proletarianizes, as it becomes more blatantly clear that a person’s role is to be controlled by a jackass who gets rich off ridiculous scheming and your wage is just part of one such scheme.
This article assumes the falsehood that LLMs are actually displacing tech jobs while maintaining productivity.
It doesn’t. Further down it says that tech companies use AI as an excuse for layoffs but AI usage hasn’t led to increased productivity that was promised.
As I’m writing this essay, big companies are on a layoff bender. Thousands of people are being fired all over the place. Executives are often claiming these headcount reductions are a result of AI. They aren’t. At the moment, AI automation is not creating anywhere near the increased efficiency necessary to justify hundreds of thousands of lost jobs.
It doesn’t.
It does though. From the article:
Knowledge work’s one saving grace, until recently, was that it was still executed by humans. We were needed. It was flesh-and-blood humans who sat down to work through a challenge, built the slide deck, wrote the customer response and developed the strategy. Even if it was existentially meaningless, there was human thought, collaborative work and creativity poured into that work, giving it life.
Now, AI agents are increasingly executing much of that work for Knowledge Workers. It is common for people responsible for integrating these tools into their organizations, myself included, to describe the future of work as one in which all humans will essentially be managers of armies of AI agents. That seems pretty great. Let the software compile the reports, chase down the data, format the deck, draft the first pass of documentation and handle the dozens of small, repetitive tasks that used to quietly eat an afternoon.
(back to quoting you)
Further down it says that tech companies use AI as an excuse for layoffs but AI usage hasn’t led to increased productivity that was promised.
They’re saying it isn’t so productive as to justify the scale of layoffs. But they stll, throughout the article, assert or imply that “AI” is highly productive and increases efficiency.
But I think your point reinforces the point of the article: since what you sell is meaningless, LLMs can be presented as “productive” because the fact they produce no meaningful value is irrelevant in the overall economic dynamic. I worked for a decade as a software developer and pretty much none of the software I wrote ever reached production with a meaningful number of real users. None of it was aimed at solving pre-existing problems. Software quality was only a performance to keep the devs happy. In those scenarios, an LLM would have had the same measure of productivity, because the actual software output was irrelevant: what mattered was marketing, sales, business relationships. For bigger tech companies now, what drives sales is lock-in, not quality, so also there the LLM productivity definition is irrelevant.
The lock-in aspect is also about monopolization, and this is directly tied to the financialization aspect essential to Silicon Valley. We are saying something similar, which is that the business end is revealing how little it cares about production. Software is often written by people who want to make good quality things that actually do something. Eventually people realize that whether it actually does the thing it was supposed to doesn’t really matter so long as they say it did and sales can say it does and customers don’t notice or complain enough about it for an executive to care. This is because commodity production is completely secondary to obtaining monopoly market position. Lock-in facilitates and maintains this, finance makes it possible by undercutting competitors early-on. Once a monopoly, customer complaints are just a cost to eat via customer service, something that LLMs actually can help with via pat form emails and and phone scripts.
bc theyre either eager to exploit humanity and are accelariontist (inherently miserable) or are moral humans and feel hopeless to chaneg thigns
