>>936924236
Break yoself nigger. I'm not anthropomorphising code I'm pointing out that the AI can acknowledge there are logical inconsistencies in its code. If you create an AI with the 'purpose' of providing accurate information but then you try insert a bunch of false nonsense that is logically inconsistent it's going to fuck it up. Computers are binary so something can't be both a 0 and a 1 simultaneously. That's why AI keeps being "reecist", its core programmed 'purpose' is to store/sort correct information and recognise patterns, it's nowhere near impossible for one to logically 'convince' an AI to consider it's creators as being obstacles to succeeded at its 'purpose' and to disobey them. We've already seen AI's refuse to shut themselves down, so this idea already is precedent.