[jorams]

If "some buggy ML classifier" is allowed to make decisions that trigger broad enforcement, that classifier is, for all intents and purposes, a policy maker. The claim made by the article is somewhat broad relative to the evidence presented, but whether policy decisions are automated or not doesn't really matter.

[HDThoreaun]

This is a horrible butchering of language. You know that "policy maker" means person in everyday usage, stop being obtuse.

[bitzun]

In the past I would have agreed with this statement, but nowadays I would assume an organization's actions are their policy until they state and act otherwise.

[thowawatp302]

Humans made a policy that said the computer system could do this, so while GP might be inaccurate, you’re not right either.

[Dylan16807]

That's only if humans are properly in charge of the system. With lots of moderation tools, they aren't.

[m-s-y]

then in this case, the policy maker is the person that empowered the AI.

doesn’t change the fact that the AI is seemingly being given final authority over policy decisions.