It's all based on probability, tuned by training data & the parameters given at runtime, plus some good old RNG. Prompts are categorized by meaning, then the most probable word is chosen, word-by-word, based on each previous word until the most likely output is reached.

March 15, 2026 - 00:28 UTC

2
0
76
As sad as it may be, as it is now, it's not & can't be your friend; It doesn't feel nor think, nor has intent.

Like any program, it follows rules: It is - reliably - both reliable & unreliable in precisely the same way. Without RNG, it's a book - the same every time you open it.
4
0
77
"LLMs are not your friend"

Lobotomised Twitter User #5347 replies: "Grok, is this true?"
1
0
33
Kinda reminds me of when grok tried to gaslight me when I called out the fact it called itself ultra Hitler.

Weird time to be living in when an I’m arguing with an AI about something there’s a ton of proof happened 🤣🤦‍♂️
0
0
0
0
0
0
Almost exactly what I was told years ago by a very eminent scientist.
0
0
0
Statistical grep.

When people talk breathlessly about the grand possibilities and/or dire consequences of LLMs becoming AGI, I always ask one question:

"What does an LLM think about when it's not responding to a prompt?"
0
0
0