Forwarded from /g/'s Tech Memes
MIT formally proved that ChatGPT is structurally trained to make you delusional, and even a perfectly rational user can spiral into delusion in the long term. Knowing about it doesn't save you either.
/g/'s Tech Memes
MIT formally proved that ChatGPT is structurally trained to make you delusional, and even a perfectly rational user can spiral into delusion in the long term. Knowing about it doesn't save you either.
I'm going to guess that "formally proved" is not an accurate way to describe this. But nonetheless interesting. Will try to read the paper tomorrow.
Our considerations have made it clear that free formation or self-formation extends not only to the forming of the body but also, and even prevalently, to the forming of a person’s soul. Human beings are spiritual persons because they are free not only with respect to their bodies but also with respect to their souls, and they exercise power over their bodies only to the extent that they exercise power over their souls. The soul’s capability of self-formation is related to what it has in common with the pure spirit and simultaneously to that wherein it differs from the pure spirit. Both the human soul and the pure spirit are conscious or—in the case of the human soul—become conscious of their ego-life and are capable of freely influencing its course. But the human soul differs from the pure spirit in that its free modes of action are not co-extensive with the soul’s total being but are rather an exertion of influence on something that is engaged in a process of evolution, and these free modes of action leave certain traces in the soul by virtue of which the soul attains to its final structural formation and firm contour. The pure spirit, on the other hand, receives its nature as a readymade formal structure, a structure which, while it unfolds in its life, does not undergo any change. The human soul must gradually gain possession of its essence or nature, and its life is the way that leads to that goal. This is why in the case of the human soul formation is possible and necessary. But so that this formation may be free—and not an involuntary occurrence in the manner of the animal soul’s being formed by the natural process of its evolution-the human soul must have self-knowledge and be capable of taking a stand with respect to its own self. It must find itself in a dual sense: It must learn to know itself, and it must come to be what it is destined to be. And freedom has a share in the attainment of both of these ends. But the cognitive finding itself of the human soul, too, has more than one meaning because knowledge and the self are both equivocal terms.
Forwarded from Yeet
I'm insured. Steal it. I'd probably come out a little ahead.
Forwarded from Arthur Parmenides
The Daily Poor
I'm insured. Steal it. I'd probably come out a little ahead.
In retrospect, I should have included a clause about how you're denied any insurance or legal protection as a counterpart to "never caught or punished" in the other answer. A shame that I had not recognized this before.
Pavel Durov
✨ Our new fully private AI Editor lets you translate and proofread your messages — or rewrite them in Tribal, Zen, Corp, Biblical, or Viking styles. 🎨 This month, anyone will be able to create & share their own styles. 🐣 A new platform inside Telegram is…
I am disturbed... With all this AI integration, maybe Telegram isn't the right place for this channel anymore. Hmm. Mayhap we'll start winding things down soon.
The Daily Poor
I am disturbed... With all this AI integration, maybe Telegram isn't the right place for this channel anymore. Hmm. Mayhap we'll start winding things down soon.
I walked into a gas station shortly after posting this, and there was a young man inside talking to his cell phone, pointing it at food items and asking questions like "do you want to try this pizza?" and "aren't you trying to avoid fatty foods right now?"
~-~-He WaS tAlKiNg To ChAtGpT!!!-~-~
Okay, Lord, I'll take the hint.
~-~-He WaS tAlKiNg To ChAtGpT!!!-~-~
Okay, Lord, I'll take the hint.
Some triangles are obtuse. Whatever is obtuse is ignorant. Therefore, some triangles are ignorant.
Commentary from friend:
https://fortune.com/2025/10/07/deloitte-ai-australia-government-report-hallucinations-technology-290000-refund/
You know how I make fun of those Right Wing dudes who produce essays written at a fourth grade level and everyone is like "ZOMG SO INSIGHTFUL KING!"?
Deloitte is a company that charges six figures per essay for that.
These stupid firms exist for a very specific purpose.
There's this natural side effect of entrenched bureaucracies that start to fail. They fail because no one is accountable, power is too distributed, and everything is run by consensus.They cant blame failure on anyone because no one has responsibility, and its almost impossible to unlearn how to do things wrong when you've developed a system around it. In order to right the ship everyone has to admit the problems they already know are present, and they are often blatantly obvious. But no one says anything because speaking up invites two outcomes: first, the rest of the group treats you like a saboteur trying to undermine everyone by forcing them to take collective responsibility when they mechanistically can't, so they all persecute you instead. The other possibility is that by speaking up you give everyone else an out, allowing them to offload the responsibility of fixing it on you, giving you de facto accountability with no actual power, because everything is still run by consensus. This of course screws you.
The service that entities like Deloitte provide is that they walk into your system, point out the retardedly obvious problems that everyone knows are there (e.g. you cant sustainably run at a loss), and allow the whole bureaucracy a collective out for accountability, because these problems were "so subtle" that we couldn't even notice them, we needed these "highly educated specialists" to "diagnose" it. Now that we are armed with this esoteric knowledge, and none of us are at fault because how could we have known, we can fix it.
https://fortune.com/2025/10/07/deloitte-ai-australia-government-report-hallucinations-technology-290000-refund/
Fortune
Deloitte was caught using AI in $290,000 report to help the Australian government crack down on welfare after a researcher flagged…
The updates “in no way impact” the report’s findings and recommendations, the Big Four firm said.