ML algorithms are already constantly deciding what you should and shouldn't be allowed to read, by surfacing, amplifying, and suppressing content on Google, Youtube, every social media platform, baked-in Windows ads I presume, etc. based on each website's own financial and/or political interests. And soon ML algorithms will be increasingly deciding what you can and can't read, by vomiting out the "content" in the first place according to those interests.

Maybe I'm just fatigued from the constant erosion of user autonomy over information consumption, but as long as a ML ad blocker runs locally by the user's own decision, it doesn't seem that Big Brother-ish (compared to serverside shovel-feeds), and as long as there's a single-click toggle to disable it like in existing ad blockers (or if E.G. the UI exposes a side pane to show the content it's blocked) and it can be configured to be less or more aggressive based on content patterns, it hardly seems like censorship… As far as uses of "AI" go.

By that logic you could call regular ad blocking censorship, because it does exactly the same thing as parent described, only without AI and less effectively. I see a difference between algorithmic feeds that I have no control over that are designed to boost the companies goals, and an opt-in tool (hopefully an open one) running locally and modifying the content presentation to remove things I don't want in the first place

Ad blocking is censorship.

Censorship that many people, myself included, agree with (I don't believe all censorship is harmful) but it is clearly censorship.

But the real problem is assuming everything an AI blocks will be an "ad." Why is this the only instance where the slippery slope doesn't apply? This is different than a normal ad blocker that uses specific heuristics to block specific elements - an AI will be employed to make "intelligent" editorial decisions based on interpreting the intent of the content. What happens when it decides - or is ordered to decide - that political advocacy or the promotion of certain ideas or contrary narratives amounts to "advertising" them?

It's literally just enforcing wrongthink, most likely using closed source centralized services run by corporations who will be harvesting and selling your data to advertisers anyway. But it's cool because ads. It's not cool when it's Nazis or doxxing transgender people or spreading conspiracy theories or plotting sedition, but it's definitely cool if Justin Whang wants to sell me on Nord VPN.

It’s only censorship if you stop other people reading/consuming the censored content.

Choosing to block ads is more like choosing to not read a book. Or skipping a boring chapter to get to the action.

Well you didn't write the filter lists yourself.

It's kind of like someone creating a list of bad books, and you looking avoiding reading any books on that list.

You choose to use the filter lists though. That's what makes it not censorship.

It's not censorship if I control the AI and can turn it off at any time.

Much more powerful entities are going to be in control. The "beauty" of AI is that it's a black box, and you can't directly check its biases. Guess who will win the people control battle? (If we are speaking of the illusion of control, it's a different matter of course)

If I suspect the AI is compromised, I'll turn it off or choose a different one.