AI Peril

I have a bad habit of telling every stupid idea or plan I come up to an LLM. On the other hand, most of the time I'm just spitting out plans I KNOW I am not going to do in order for them to stop circling my head.

I have always been wary of LLMs, and I never joined the hype, always being suspicions, and through experience, I've known them to be worthless. Yet, of all my friends (admittedly not many), I am the one who most uses LLMs. Perhaps, it is because I have no-one to talk to, having substracted myself from most of the internet at this point.

I usually write in order to "think out loud", so to speak.

Sometimes I use an LLM to try and flesh out some idea for an essay. In the prompt, I actually outline the structure and the main points of my argument. Only to have it then drowned by the LLM's "feedback": a flat wall of text written as if it were a reddit post or a blogpost by some middle-class white lady who sucks at writing for a living.

LLMs, social media, doomscrolling, and "the algorithm" have all had a positive effect on me, however: not using them. The negative effects of all of these have been the subject of much talk lately. And because I refuse to let my brain get deteriorated by these, I have sought to push against them however I can. I actively try to train my attention span, my problem-solving abilities, try to think through ideas on my own, and even learn to draw so that I don't find myself trying to realize a picture through slop.

If "urbanites" are willing to let their brains become jelly, I am going to make myself actually useful by developing the skills that many are striving to lose.

On a somewhat related note, I considered substack for a while. Perhaps I have something interesting to write such that I can grow an audience and perhaps even make some money off it. Then I learned that substack apparently works like most social media, and I don't want to be at the mercy of some algorithm and subject my thoughts to the degradation that the free market inflicts on everything it touches (I really have to write that article about how the free market deteriorates the quality of things, soon.) I would rather *pay* to get my writings out there than pretend to earn money by doing so. Which is more congruent, after all, with my own notion of how "economics" should work.

I am not a luddite (not completely, at least), and I do not push against technology just for the sake of it. I have found there may be a few times when LLMs are somewhat useful:

• Review my writings. I passed a couple things I wrote through an LLM (Claude, specifically) for review, and they were kind of useful. I might use it for that purpose, sparingly.
• Quick rundowns. There are some times where I am just not going to spend an hour researching a topic. Sometimes it's more straightforward to ask the LLM for a topic. Most LLMs today do a web search before answering, and I will often go right to the sources they cite in order to check for myself and also to see where those sources come from. They may be financed by anglozionism, after all.
• Deepwiki. I just found out about deepwiki. It seems to do a decent job of scanning through a project's code and outlining it's overall structure and function. I can use it as an entry point into the inner workings of some codebases I'm very interested in.

That said, I need to make sure I don't end up depending on it where my brain should do the work, nor believing the machine's word without double-, or even triple-checking.

Alternative

I said I often don't have anyone to talk to. Talking to an LLM, however, is just like talking to a wall of text, which bounces back what you said in a convoluted, generic manner.

I have found a better interlocutor, or companion, who gives far better answers than any LLM could ever, and that uses far less energy, and pollutes a lot less water.

It's the Tarot.

I'll leave the verification of this as an exercise to the reader.

Until nex time, when I will talk about the free market or zionism.