LLM circle jerking
Let me use a viral image for once, yeah?
I made the mistake of browsing LinkedIn after my 2-week holiday, and boy, was I absolutely ASSAULTED with "oh my god, AI is so useful", "oh my gawd, AI made me a 10x engineer", "omg, AI bla bla blaaaaaaaaa". I cannot read these posts without making all sorts of harrumphing sounds, snorting, coughing with disbelief, rolling my eyes. What are these people hoping to achieve by posting this absolute big tech cocksucking drivel? Internet points, that's what. It's the currency of our time that so many of us are addicted to, because Big Tech has successfully hacked our psyche.
Funnily enough, LinkedIn then gives me the option to post automated comments like "100% agree", so I can join in the circle jerk without having to type a word. How convenient.
I cannot cross the wall of cognitive dissonance, dear reader. I think LLM's are mostly crap, and I can no longer be bothered to use them. Back in 2023, I had a brief period where I tried out ChatGPT Pro, and it made my work experience worse. It dehumanises the work I do, and this might sound shocking to you, but instead of trying to be the most productive capitalist little slave I could be, I actually want to have a pleasant time doing actual useful work. This work involves using my brain, and this cannot be offloaded to a machine. (Also, perhaps because of my work as a software tester, I have a higher tolerance for repetitive work. I always try to find novel ways of looking at things that should be done repeatedly, exploratory thinking rules, y'all.)
I am no longer writing code in my day-to-day work, so I have no need of offloading repetitive work to an LLM. I think, I research (with Kagi search engine, not an LLM), I write, I present, I discuss, I coach, I work with people. None of these things benefit from an LLM.
In my spare time, I am an avid consumer of LLM sceptic blogs. So yeah, I'm in the circle jerk on the right. Whenever I read things about how much LLM's suck, my confirmation bias purrs like a content kitten with a belly full of milk.
I wonder who is still capable of crossing the wall of cognitive dissonance. Who is not either a fan or a sceptic?
What even is the middle ground here?
Even if you say that "w3ll aCtuALLy, LLM's can be useful in certain use cases", you are still glossing over other facts: the big tech companies that provide you with those LLM's only want to earn more money, they stole everything on the internet to train their shit, they want to remove your job, they don't give a fuck about you! And what about the fact that the energy usage of LLM's is atrocious? When it was crypto, everyone was pissed at how energy slurping it was, but now those same people are happily using "chat" as their therapist. Fuck right off, please.
So yeah, if you consider yourself part of neither circle jerk, please enlighten me how you do it. I want clear examples of how you are able to strike a middle ground here, and how you reason away the general issues that lie at the core of this technology.
Because I am not seeing it, I cannot move beyond my cognitive dissonance.
PS: I made these images in Canva, marvel at my powers.
I have just finished reading the book "Blood in the Machine" by Brian Merchant.
Brian Merchant
I should write a separate blog about the book, but it's going to take some time to compose my thoughts. If my procrastination wins, the post might never come.
TL;DR: we have been in the same cycle since the 1800's: at war with the machines, capitalism, and entrepreneurs, and AI/LLM crap is just the next wave in the same cycle. Read it, and be enlightened by the power of history. We simply never learn from history, that's the problem.
Doom metal still rules
The new Paradise Lost song slaps. It reminded me of how much I love the subgenre of doom metal. Tickles my brain in just the right places.
Comments ()