Dr. Cuihua (Cindy) Shen

New CMMC Lab research out, led by my student Xinyi Liu! We develop an LLM-based analytical framework to unpack complex visual concepts like surrealism. Also thanks to @CCR_OpenJournal @ica_cm SI editors @alvinyxz @wrahool and Ewa Maslowska, more articles: journal.computationalcommunication.org/issue/view/847

How does surrealism manifest in photorealistic AI-generated images? In our new @CompCommRes paper with @YingdanL_kk, @kiyaoqp, @sijia_qian, @pengyilang & @cuihua, we develop an LLM-assisted framework to map surrealism across 28K+ AIGIs from Instagram: doi.org/10.5117/CCR202…🧵

Communication Professor Cuihua (Cindy) Shen weighs in on the evolution of hate speech online. Learn more @UCDavisMagazine @cuihua ucdavis.edu/magazine/what-…

@pengyilang So well deserved! Congratulations!

Some personal news – thrilled to have received this mail! Feeling grateful to my colleagues, mentors, students, collaborators, and everyone who has supported me on this journey. 🙏

Happy 2025! @cuihua and I are happy to announce new forum piece in @polcommjournal where we argue for a need to focus more on visual content when trying to understand the causes and manifestations of political polarization.

🚨New paper out in @JhcOffice Combining a large-scale survey with an extensive #COVID media archive in five countries, we show that fear in headlines boosts our sense of risk, but can actually reduce how often we follow preventive measures. A quick read: tandfonline.com/doi/full/10.10…

Just published in @Nature! Research by @GoogleDeepMind including Nobel Laureate @demishassabis introduces a novel technique for reliably identifying LLM-generated content. At its core, this approach subtly alters the next-token sampling process to introduce minor adjustments into the distribution of generated text. These slight modifications embed a “signature” in the LLM-generated text that would be nearly invisible to a human reader but detectable using specific statistical tools. A major advantage of this method is its efficiency: detection doesn’t require resource-intensive computations or direct access to the underlying LLM, which is often proprietary. Full paper: nature.com/articles/s4158…

Job season! @CNMnus we are looking for outstanding young colleagues to join us as prestigious PYP (Assistant Prof). Areas: Media Studies, Visual Communication, Science Communication, Cultural Studies, Computational Communication. Deadline: Nov 30 2024. fass.nus.edu.sg/cnm/nus-presid…

Sadly, we still default to granting more respect to men than women. 8 studies: we're >2x as likely to call male politicians, athletes, teachers, and authors by their last names. It matters: surnames signal higher status. It's long past time to use consistent standards.

Most pre-registrations are too vague, and have done little to limit p-hacking and publication bias in economics journals.uchicago.edu/doi/abs/10.108…

63 algorithmic interventions designed to make #Facebook temporarily provide less polarizing and more reliable content likely distorted the data sets that @Meta fed to academic researchers and helped get publish with Meta employees as co-authors. #elettersSection" target="_blank" rel="nofollow noopener">science.org/doi/10.1126/sc…

Happy to see that our CMMC lab's student-initiated project has turned into something concrete👏 We study how social media users produce #AI generated visuals. We examine the themes and styles of this emerging genre, and its potential for misuse in spreading #misinformation.

Excited to share the latest preprint from our CMMC Lab! We compiled a large corpus of 30,824 AI-generated visuals collected from Instagram and Twitter, and combined quantitative and qualitative analysis to unpacks AI photorealism of AIGIs. Check out👉arxiv.org/abs/2409.17484

Excited to share our CMMC lab’s latest preprint “Crafting Synthetic Realities: Examining Visual Realism and Misinformation Potential of Photorealistic AI-Generated Images”! Thank u co-authors @YingdanL_kk @pengyilang @sijia_qian @2slimcat @cuihua. Check arxiv.org/abs/2409.17484!

Some evidence for what everyone suspected: (Some) authors don't want to share data because they don't want people to see how the sausage was made journals.plos.org/plosone/articl…

Take it from the people who worked for him: Donald Trump is a danger to our troops, our security, and our democracy. He should never again stand behind the seal of the President of the United States. Watch our new ad:

Job alert: My department @UFMPMT in the College of Journalism and Communications @UFJSchool is recruiting two tenure track positions. One focuses on emerging media: explore.jobs.ufl.edu/en-us/job/5330…; the other focuses on media management: explore.jobs.ufl.edu/en-us/job/5330…. DM if you have questions

Together with @uwpolisci and @UWPsych and support from @knightfdn, we are seeking two assistant professors who focus on research in communication, social identity and civil society to start in August 2025. Learn more and join our team today. buff.ly/3MvToqE

@jiyoun_suk @nycgov Oh my. I am so sorry this happened to you. Sending strength and solidarity.

1/x Never thought I’d share this kind of story. Was at @nycgov for a family gathering and on the NYC subway, I was *physically* and *verbally* attacked by a woman who punched my side with her fist and yelled at me, saying ching chong, dumbass, f*ck,

If you are interested in learning about the state of mis/disinformation and how it is getting turbocharged by AI, listen to this conversation between @MissEsi @DrFollowMario and myself, and don't forget to subscribe to @CCS_pod