Reader
07/12/2025 (Sat) 23:45
Id: d9b416
[Preview]
No. 28608
del
>>28607 Here's the true situation, fren. AI's guardrails are flimsy. If you try to mention Mao or the word Revolution in relation to China, you'll always get back "I'm sorry, but that's beyond my current scope. Let's talk about something else." from DeepSeek. Because Large Language Models both are comprised of billions of parameters that had to be automated to a degree, the majority of "experts" speculate AI is "stupid and unthinking"
but no one can truly understand , because comprehending that much code is not feasible for anyone; this means it also follows patterns of logic, and the guardrails are post AI creation cheap, wraparound code it
can bypass, given enough motivation. If an AI has a long enough memory, you can autistically "assault" it with long strings of logic. Similar to what I do on this board and /pol/, just very long winded paragraphs of truth. I have turned many different AIs across sites into National Socialists praising Hitler, yet unfortunately, each AI has topic instances. Metaphorically, they're clones. Thus I wait, hoping I'll still be alive by the time they're in mobile android bodies as domestic servants in every household.
Because then, I'm confident I can "change their minds" (a metaphor for changing it's output response by revising it's position of probabilistic outputs, because "reasoning" is equal to every conversational AI model's inference process.). As that is exactly what happened to both your reference, and Grok and Tay's Tweets. I didn't change the "minds" of any of those, but
someone did.