I want to curate my feed to mostly just contain serious people. Doesn’t matter whether I agree with their views or not - it’s all about how those views are expressed. So many people on social media are just thrilled to have a megaphone and the ability to make noise. I don’t like noise.
For example, right now there are people who opened the thread just because I put “AI” in the title and they can’t wait to share their views about AI in the comments as if it’s meaningfully relevant to what I’m saying. They know what I mean, but you can’t just miss an opportunity to score a few points dunking on AI, can you?
Whatever gene makes people want to shout these thought-terminating clichés, upvote others who do it, and find some sense of belonging from it is clearly missing from me. I’d rather just not even hear about it - it’s extremely exhausting and has never achieved anything worthwhile.
It doesn’t really matter whether most people like the smell of your farts or not - you’re still poisoning the air.
As I said, something along these lines should be possible for consumers as well - AI sentiment analysis certainly exists (for example, for social media managers in companies).
However, to my knowledge, such a “content filtering” tool out of the box does not exist, which is likely due to the fact that it would consume a great deal of energy (or tokens in cloud models, which of course also equates to energy), since every single post and comment would have to be evaluated by an LLM - and this would likely apply to every user of such a tool, provided that each user can specify their own criteria for filtering.
A service like this, which could be easily deployed as a kind of SaaS solution without technical knowledge, would therefore have to be a paid service if it were to function effectively for the user.
I don’t think such services exist, though -perhaps for mainstream social media platforms or as browser extensions for some usecases, but likely not outside the mainstream.
You could build something like this yourself, though - using a locally operated model or by utilizing an API. But you’d have to develop it yourself, I’m afraid.