[abusoufiyan]

Few things I realized learning AI/ML in school and being around people who claimed to care about "social good" and "ethics":

1) As long as engineers chase after money and prestige first, there will be no ethical AI/ML. We live in a society which values money and status over everything else, especially moral values (it's quite trendy these days to reject traditional moral values, in fact). If that doesn't change, the incentives to create unethical, untransparent and unfair AI/ML don't change either.

2) The trendiness of ethical / fair AI/ML is the only thing it has going for it, and if companies can coopt it while actually using AI/ML for unethical goals, they will do so happily (for instance, if they were to sponsor ethical AI conferences or papers while still using AI to spy on people and draw up lists of possible terrorists without hard evidence, ala Palantir).

3) There is no desire to have a hardline, this is right and that is wrong code of ethics like doctors do with their Hippocratic oath and a large part of this is the entanglement that a lot of AI/ML research has with the military. It's not possible to talk about ethics and ignore the fact that lots of this AI/ML research is going towards killing over 1 million innocent civilians in Iraq. And yet, nearly all the labs which do this research receive US military funding. When this is the case, you have to leave loopholes for people to reason their way into justifying things which are, to most people's base instincts, morally wrong. And that leaves a kind of ethics where everything is ultimately permissible provided you think about it hard enough, in other words no ethics at all.