Yep, we've been using Langit: github.com/rimar1337/la.... The client has changed quite a bit since we first started though.
fun test fork of a Opinionated web client for Bluesky social media - rimar1337/langit
Yes exactly! We've actually forked an existing bsky client to implement a very experimental version of the natural language feedback idea from our paper and have done some user studies with it. Hope to share the results in the near future!
Concerned about what’s real vs fake, esp after seeing Sora videos? In a new blog post, I (w/ Jina Yoon & @axz.bsky.socialwww.cip.uw.edu/2024/02/21/p...
While provenance can indeed help people better calibrate trust and accuracy perceptions of media, it can also come with its own risks of misinterpretation
Cool!! Just curious—how is the total # of feeds being counted?
Meta is rolling out labels for genAI media! See our recent paper studying the design and impact of provenance indicators - we found they help ppl spot deceptive media w/ some caveats. w/ @kjfeng.bsky.socialdl.acm.org/doi/abs/10.1...about.fb.com/news/2024/02...
Nick Clegg offers a new approach to identifying and labeling AI-generated content.
Ah ok, I understand. LLMs are a part of the experimental prototype we developed, but the controls we're testing don't necessarily require the use of LLMs in the end. If you prefer not using LLMs at all, we're planning some future studies that don't involve LLMs, so keep an eye out for those!
The tool we built has an LLM-powered feature but you don't need to know anything about LLMs to participate!
I participated in an interview with a group of fantastic UW scholars of social media for UW News! 🤩 We talk about problems with large, centralized social media and whether the fragmentation we're seeing of today's social media landscape has been beneficial: www.washington.edu/news/2023/10...
The turmoil at large tech platforms has many people reconsidering what they want out of social media. Four researchers at the University of Washington are exploring different approaches to improve...