LLMs are a mixed bag. I judge them case by case, and the pattern-play technology strikes me as negative overall due to the environmental costs, error-prone results, displacement of human beings with no plans to help them, and its ridiculously overhyped status due to these companies needing to regularly raise billions of dollars to beat the world into submitting to their profits-for-them vision of our future.
Oh, and one other thing: The social contract underpinning work is being destroyed so a few more well-off people can make a lot more money.
Does anyone remember that we are a society of people presumably trying to work together to survive and thrive? Do we understand that societies can collapse into senseless violence when nothing holds them together?
Do we know that communication is sacred? That it is what makes us human — that we create our humanity every moment through language?
Pressing a button to outsource messages to our fellow human beings (it’s so pedestrian of me to not see them as flesh-based profit centers to manipulate into maximum ROI!) is the most dystopian, insane, and suicidal decision I’ve seen being pushed by hypercapitalism’s club of power-hungry sociopaths who will never be satisfied by anything or anyone.
A rare detour from the overall spirit of this rant: For some reason, using LLMs to crunch huge piles of human-averse data feels less offensive to me (protein analysis, e.g.). If machines can accurately track thousands of variables at once and identify how they affect each other, that’s pretty interesting, though of course it comes with the usual questions around employment and the role of human analysis. It’s the use of a bot to replace communication that tends to set off every alarm bell I have left (most rang themselves to death starting in 2016).
But, of course, I’m just being a reactionary. I’m a Luddite. I’m a fear-driven, tech-hating monster who is simply too immature to embrace the bright tomorrow our tech overlords will deliver if only I’d shut up and get with the program. Etc.
Our society’s god is Efficiency. Anything humans do to demonstrate their humanity is inherently inefficient. To increase efficiency, humanity must decrease. We now have a nuclear weapon to decrease our humanity at scale.
Gosh, it’s so exciting and future-y, Forbes had to run a special series about it, and I had to read each article to know the latest about LLMs. Enjoy the epitaphs:
- Diagnosing a fleshsack’s health cheaper, faster, efficiency, more profits, blah, blah: Veterinarians scored numerous videos of cows walking to give the LLM patterns to look for and score: https://fortune.com/2025/06/26/ai-farming-cows-cattleeye-health-environment/
- How-to chat manuals: A financial firm feeds a pattern-recognition bot a bunch of info for it to then regurgitate in a chatty way. Also, the draw of LLMs is their general-purpose utility, allowing this firm to abandon or to consider abandoning more bespoke software solutions they’ve built over the years. So: It is, at the moment, a chat-interface version of a how-to manual for the staff; let’s hope it doesn’t confidently fabricate information, because no one is going to spend their day checking the source material, since that defeats the purpose of using the fabrication-friendly, pattern-recognition machine in the first place: https://fortune.com/2025/06/26/ai-banking-bny-eliza-personal-information/
- Programming other machines: Correctly programming a robot to machine parts is a complex process involving a zillion or so variables. A pattern-recognition bot churns through them to craft instructions that humans review. Interesting twist: the human operators of CNC (computer numerical control) machines don’t share their knowledge, so individuals across firms and geographies operate in isolation. This industry allows no room for error, so the idea implied is that these folks have figured out a way around the “hallucination” problem of LLMs. It is not described beyond saying a human operator reviews the LLM’s output. https://fortune.com/2025/06/26/ai-manufacturing-parts-cnc-machines-american-industry/
- I am sure employees will love to experience the immense respect shown to them when their employer can’t even bother to talk to them in person: A machine makes videos of LLM-generated copies of people talking about whatever. Blah blah, you can now program a bot to talk updates to your sales team, blah blah, now a bot can do the weather or walk your employees through the HR team’s complaint process, blah blah, so efficient, blah blah. https://fortune.com/2025/06/26/ai-digital-avatars-corporate-video-training-gen-z/
- Okay, maybe this “best sorting device ever” angle on LLMs is good? Recycling fails because filtering a pile of waste is hard and slow. Robots with LLM-vision can do it faster. This seems good? But maybe someone should mention the LCA (life cycle analysis) issue with recycling? Maybe we should not try to keep poisoning ourselves with plastics and instead eliminate them from as many goods as we can as a starting point? What silly ideas. That would mean fewer profits for oil companies. https://fortune.com/2025/06/26/ai-recycling-trash-amp-robotics/
- If robots are the only ones who know the law, the robots are the law: Knowledge is power. But power is human. And humans’ relationship with knowledge shapes their relationship with power. If knowledge is no longer earned through physical and mental strain, the actions decided by knowledge will be come violent, detached, and anti-human. Remember, nothing is basic the first time you do it. Take away that learning, and there will be no one around to approve the accuracy of an LLM’s vomit. “It’s been obvious for some time that AI is a good fit to displace large swaths of mundane ‘grinder’ work—and indeed that is occurring with tasks like document review, contract drafting, and basic research.” https://fortune.com/2025/06/26/ai-law-legal-practice-attorney-client-relations/
- A world led by sociopaths who have fully accepted the hype. It is somewhat satisfying for the truth to emerge: “A recent IBM survey of 2,000 global CEOs found that while leaders are still confident that AI is pivotal to their future, only 25% of their current AI initiatives delivered the returns on investment they had hoped for.” https://fortune.com/2025/06/26/ai-at-work-companies-return-on-investment/
We have stopped being a society. We are a machine destroying life to make magic green bills. All of us are individual microcosms of that macrohorror, and now the robots are coming for the last few lines of our human poetry. A race to oblivion for TVs, cars, and fridge cigarettes. Self-destruction used to take some work. Now it’s automated.
I initially posted this on Substack as part of an ongoing experiment. I suspect that service is suffering from its own success and needs to prioritize revenue-generating authors over newcomers who contribute sporadically sans a corporate business plan for their one-person prayer for income in a collapsing economic order.