Muh 'machine fraud is to complicated, we can't trust normies to understand, there's only one type of fraud that matters, don't look over there' revolver/ gateway pundit type pieces
>>20141540, >>20141541 @elonmusk - Don’t Look Up (ChatGPT Created an Image of God with a Bunny)
Parlor trick. Ai knows what it is given. It is given human at. It is your what words are associated with what type of art. The bulk of 'philosophers' on the Internet who are using the word 'transcendent' for instance, are also likely to associate such words with their mushroom cult. Ai obeys whatever associations prolific Internet publishers (teenagers) dictate.
Stop letting programmers convince you machines have minds of their own. Make them prove it by showing you the insides.
https://open.spotify.com/track/5YIh8uixmY4g4cC1MxoXnt?si=vQGWNuMATsK3Dq0ckctPxw
They beat it through brigading. 'reputable' sources all cite each other. Same way peer review gets gamed. All the most 'credible' sources say 'baseless' and ask if a sudden you have the crowd repeating it and it becomes a self fulfilling prophecy. Ai sees everyone in 'consensus' and automatically down rates outliers who are rational free thinkers.
There are many ways this could be true that would require elegant packaging to be understandable (data). If it is otherwise 'impossible to refute' on first glance, it seems difficult to rationalize waiting.