Feature request: Suppress number of AI influencer ChatGPT tweets

[NielsRogge]

Is your feature request related to a problem? Please describe.

Hi, thanks for open-sourcing the algorithm behind "For You".

So recently, a lot of people in the AI community (including myself) were a bit frustrated by the fact that a lot of tweets containing "here are 10 things you need to know about ChatGPT" followed by the 🧵 emoji are favored a lot, and cause a lot of noise in the users' Twitter timeline (especially people from the AI community, i.e. AI researchers, ML engineers which heavily rely on Twitter to share and discuss content).

There was also a meme created for it (source: Hardmaru, a famous AI researcher who was formerly at Google Brain, now at Stability AI):

Here are some example tweets that pop up in my feed frequently which have similar content:

• https://twitter.com/heyrobinai/status/1659960933878300673
• https://twitter.com/TheRundownAI/status/1659985094453473283
• https://twitter.com/aakashg0/status/1659419606078066688

The trend is actually not limited to the subject of ChatGPT, it seems that tweets containing "here are ... you need to know" and containing emoji's like 🧵 and 👇 have a higher chance of being displayed/recommended, as they obviously are a bit clickbaity. I was wondering whether the Twitter team could look into providing an option to the user to suppress these kind of tweets/recommend them less.

Describe the solution you'd like

It would be great to provide an option to the user to suppress those kind of tweets or recommend them less. Currently, there doesn't seem to be an option available like in other social media apps (Facebook, Instagram) where you can select "I'm not interested in this kind of topic" or "show less tweets like this one".

Describe alternatives you've considered

As an alternative option, one could mute tweets containing the words "here are 10" for instance, but this is not an ideal solution as it's rule-based and of course, an embedding-based approach would work better.

Thank you for considering!

Niels

[ngngardner]

You can mute the emoji, that's what I've done

[TolstoyDotCom]

Keeping famous AI researchers from being annoyed is one of the world's top concerns, but as for me I'm more concerned about how Musk - just like Vijaya before him - is greatly helping the leaders of dictatorships hide from dissent. For instance, Twitter heavily censors replies to Chinese and Iranian officials.

Not to mention the millions of other users they censor each day (with most of those users being blissfully unaware of it).

Instead of trying to make your completely unimportant issue important, it would be better for you to support eliminating "The Algorithm" entirely or at least make it optional. As I and others have suggested, Twitter should let you filter out things you find annoying, but your filtering shouldn't impact other users in any way.

[NielsRogge]

Lol even Elon Musk tweeted a meme about it now, and then later deleted the tweet

Here's a screenshot of it: https://twitter.com/still_oppressed/status/1664813099218776064/photo/1

[TolstoyDotCom]

This is a completely trivial issue, the only aspect that's of use to society is that it illustrates yet again how technically incompetent Twitter is.

Here's an actually important issue: Musk's Twitter is heavily censoring replies to Chinese govt officials. See the 2nd tweet. The empty tweets that are in the "Anomalous elevated replies" section are probably childish GIFs or something.

The important thing is the tweets in the "Anomalous suppressed/hidden replies" section. The ones in red are marked by Twitter as "LowQuality" and those in purple are marked as "AbusiveQuality" (by the beyond-flawed algorithm in this repo).

Musk does something similar with replies to the head ayatollah. I'd say that's a lot more important than some very special snowflakes being annoyed. Maybe they can get a prescription to deal with that issue or something.