They created 323 “sock puppet” accounts—fake accounts programmed to simulate user behavior—across three politically diverse states: Texas, New York, and Georgia. Each account was assigned a political leaning: Democratic, Republican, or neutral (the control group)...
To analyze the political content of the recommended videos, the researchers downloaded the English transcripts of videos when available (22.8% of unique videos). They then used a system involving three large language models—GPT-4o, Gemini-Pro, and GPT-4—to classify each video. The language models answered questions about whether the video was political, whether it concerned the 2024 U.S. elections or major political figures, and what the ideological stance of the video was (pro-Democratic, anti-Democratic, pro-Republican, anti-Republican, or neutral). The majority vote of the three language models was used as the final classification for each question.
Imagine believing one super sketchy study from some dudes in the UAE ffs. Even if this is accurate (highly doubtful), it might be because kamalacaust didn't campaign in swing states for example. There's no real evidence that this is real and, even if it's real, there's no evidence that's it's due to bias at TikTok.
However there is plenty of evidence that people are trying to ban TikTok because it reveals unfiltered information about palestine, etc.
Who loves slavemasters? Founding daddies? Free from religious control?!? These creeps were the epitome of unchecked racist power. It's hard to imagine that adults believe/upvote this wacko nonsense.
Not to mention the past 100 years of palestinian life.