Where's the AI angle?
An algorithm that delivers the wrong answer is not unethical, it is simply wrong, whether or not it is using what decades-old-algorithms are now being described as "AI".
Conversely, a company that blindly uses whatever comes out of their computer to guide legally actionable decisions *is* unethical, not simply wrong. Once again, this is irrespective of the sexiness of the algorithms involved.
Biting the hand that feeds IT
