>but actually the average person doesn't want technology to vocally support neo-nazis? By papering over these flaws in AI without actually fixing them, it comm

>but actually the average person doesn't want technology to vocally support neo-nazis? By papering over these flaws in AI without actually fixing them, it communicates to the average person that the technology built is more sophisticated than just parroting whatever nazi propaganda it picked up online. I think that's fundamentally dishonest. The level of distrust in AI that the average person *should* have is the same level of distrust they'd have if it also randomly supports neo-nazis.