We have all been inundated with posts and stories about LLM's and AI of late. I have attempted to somewhat calm the furor by posting a factual description of what these devices do, which demonstrates they aren't in any way intelligent. They're just algorithms, mathematical statistical analyses of content designed to weight potential extension of the prompts they are given. Reduced ad absurdium they are basically seesaws that respond to user supplied weights.
However, it is a fundamental aspect of human nature we project our own mental landscape into the heads of others when considering their actions, and we naturally project our ability to think and reason into text authored by LLM's. Only reason enables us to reflect on our instinctive responses and moderate our conscious actions thereby, and I think we can agree that observation of the world today provides ample evidence that reason is in short supply.
We are all aware of the predatory nature of financial entities. Pick your poison, whichever Big Tech monstrosity you choose has been demonstrated repeatedly to criminally harvest PII and monetize it, M$, Goolag, Appfail, whatever, doesn't matter. They are all demonstrably criminal organizations and so focused on their aggrandizement they are malevolent. If it is profitable to do harm, they do harm.
LLM's and all AI are products these corporations are devising to aggrandize them. Goolag and it's ilk aren't striving to advance human felicity. They're striving to profit. The WEF is a forum in which the most powerful of such organizations advance their interests, and it clearly lies behind international policies that are degrading national governments globally today preparatory to implementing a NWO that aggrandizes the infestor class, the majority stockholders in the most predatory stock corporations in the world.
The global uniformity of plandemic policy that has no basis in scientific fact is proof that national governments are ubiquitously corrupted by the universal mechanism of financialization. The FDA isn't a captive agency. The USG is a captive mascot of oligarchs.
I suggest reframing your consideration of geopolitics by this understanding. It all makes sense then.
Yet we have failed, largely, to note that AI is typical of the products of such Big Tech corporations regarding profiteering from PII, or any other human attribute that can make corporations money, whether it's our content, our purchases, or our health. If there's an angle corporations can take that profits them, we can ubiquitously observe they take it.
So it is no surprise to me to note that meltdowns of LLM's have resulted, because I would not expect products of these corporations to be created for any sort of benevolent purpose. These products are made for the purpose of aggrandizing infestors in the corporations that produce them.
I note the following comment to the article referenced.
""It was phrase-extending and it hit a gradient in parameter space that took it down a non-factual path. "
"And someone gave it that "parameter space", it can't invent links out of thin air and claim they are true. No amount of fuzzy logic can create links."
These products aren't making mistakes. They're designed to do harm, any kind of harm that can result in profit to the corporations that produce them. The AI can't lie, because it has no basis to differentiate fact from fiction, but the programmers of the AI can, and if lies profit the corporations, then lies are programmed into the AI.
LLM's can't lie. They can't create text out of the blue, but only weight examples they were trained on. These lies are what have been programmed into them by the corporations that produced them. These aren't only LLM's. They have been tainted with malevolent programming that will profit corporations by harming you, me, and humanity.
If you're using these tools thinking they benefit you, think again. You're wielding a weapon intended to be used against you - by the pointy end.