Most of the AI horror scenarios are based on it applying things literally and without regard to consequences. In fact, there’ve been some arguments that a conscious AI would be less likely to fall into that behavior than a zombie, because it would have a theory of self.”
