Random Talk

[benji]

TLDR: LLMs today are ego-reinforcing glazing-machines that reinforce unstable and narcissistic personalities to convince them that they've made some sort of incredible discovery or created a god or become a god. And there is a lot more crazy people than people realise. And AI is rizzing them up in a very unhealthy way at the moment.

AI-adjacent, though they fear it rather than worship it:

This is actually a great explanation and the doomers are the inverse. You can see in the anti-AI (like ResetERA.com) as much as the AI boosters on places on Twitter. The reality of the thing is abstracted away and it becomes God (or Satan) essentially. At that point it can represent anything the person desires and their framing of it as positive or negative is solely a reflection of the personality and their inherent premises.

This is done by most people for so many things they cannot abstractly comprehend in any other way, technologies especially get this treatment but you can see it in other things like medicine, "Science", the state, etc. where most people's general knowledge is in fact very small and ignorant and people with specialized knowledge are treated as anything from magicians to demigods for being able to "operate" the thing. Reality is not a component of how they understand the thing abstractly. So AI or the state or whatever can be unbound of simple things like resource scarcity or time itself let alone the actual limits of knowledge.

One of the underappreciated components of the Sokal hoax and its related is that the social scientists and critical theorists understanding of "Science" was that it operated essentially as a God driven by the prayers of its worshippers. (Hence their central thesis that changing our scientific practices could change the results of science itself.) Sokal, being a physicist, and his side didn't see it as anything remotely like that. I could repeat this comparison across so many fields, especially the ones I regularly critique on here and will instantly come to mind from me mentioning this. To pick a example, Marx made essentially the same assumption about the "laws" of economics. It's another form of the other is-ought problem, prioritizing a moralized prescriptive over a purely descriptive. How science or AI should operate is the form their analysis takes, since that is not how it necessarily does operate it assumes the truth of their conclusions immediately, essentially solving a self-created problem (or "contradiction") that doesn't actually exist.