51
TikTok users don't have as much agency over their FYPs as they think (arstechnica.com)
submitted 3 months ago by to c/world@lemmy.world
4 comments fedilink hide all child comments
 

cross-posted from: https://lemmy.world/post/49222458

Surprise Surprise ;)

"But some users have voiced concerns that TikTok’s almighty algorithm doesn’t seem to incorporate negative feedback very well. Even when they don’t watch a suggested video or click the “not interested” feature, they keep seeing those videos on their FYP. Northwestern University computer scientists put those suspicions to the test. According to their recent paper, the engagement signals do have an effect, but only temporarily. Then the algorithm gradually relapses unless a user consistently gives the same feedback over and over again."

all 4 comments

sorted by: hot top controversial new old
[–] 25 points 3 months ago (1 child)

FYI

TikTok’s For You Page (FYP) is the default home screen for users of the video-sharing platform

So I guess it’s like your main feed.

  • permalink
  • fedilink
  • source
  • hideshow 1 child comment
    •
    • [–] 0 points 3 months ago* (1 child)

    Their methodology did not involve computer simulations; rather, they created bot accounts on the actual TikTok mobile app, rather than studying actual users. “We used emulated devices, where we are creating accounts and automatically interfering with the TikTok algorithm through code with the sock puppet accounts,” co-author Levi Kaplan told Ars. “We’ve come up with a methodology where we get the metadata by intercepting the network traffic, and then we make a decision using an LLM. All the LLMs were validated with human responses as well.”

    Super reasonable right? Definitely not buzzeor bullshit with a healthy dose of non deterministic probability tossed in there. Reasonable right?

  • permalink
  • fedilink
  • source
  • hideshow 1 child comment
    •
  • [–] 10 points 3 months ago

    I'm having trouble parsing this. Looking at the paper (https://ojs.aaai.org/index.php/ICWSM/article/view/42688) it looks like they're using the LLMs to determine if a video aligns with one of the three categories they were testing this behavior against (cooking, fitness, and sports betting). They decided to go with that rather than hashtag filtering because hashtag use on the platform isn't moderated. So, unrelated hashtags can be used on a video, some don't have hashtags, and none of it is standardized. Categorizing things is something that LLMs are pretty damn good at, actually. Mostly because we actually understand how they do that (localization in high-dimension vector DBs)

    Like, they're not using the LLM to interpret their results, man. This is a clear centaur, not a reverse-centaur. The tool is being used by those who understand it, not being made to make decisions that it has no business making.

  • permalink
  • fedilink
  • source
  • parent