Actions
(78.48 KB, 1050x1870)
>>/33478/ yeah nah whatever, GPT-3 talks like that because it was fed material that had these ideas, the model just rephrases them / re-frames their context, those articles it generated presented conflicting ideas too probably unless 'not destroying humanity' was beaten into it's head with training material and won't output anything antagonistic unless randomness is cranked up I talked to GPT-3 long enough through AID to know that it doesn't have any coherent thought patterns except for craving for cock