Our latest JCMC paper explores the emotional implications of AI companionship by analyzing 35,000+ posts from r/replika. We uncover seven types of human-AI interactions and their links to user emotions, revealing the bittersweet "paradox of emotional connection with AI". 🤖❤️
Got hacked by scammers and bots in a recent online survey after recruiting from Reddit. Over 4,000 responses became just 200 valid ones after a major data clean-up. These digital tricksters were craftier than anticipated. It's a new era of online scams (1/9).
Proud supervisor moment with my first PhD graduate! 🎉 Congrats to
@zichengstephy
for excelling in her PhD journey. Now a postdoc at
@NUSComputing
, she continues studying digital well-being. Big thanks to co-advisor
@lexaaylexaay
and
@CNMnus
for invaluable support and guidance!
Congrats to the team! Our
#CHI2024
paper analyzed nearly 1M posts on Twitter and Weibo to explore public discussions of conversational AI across cultures. The results highlight the importance of designing AI chatbots with cultural sensitivity. Preprint👉
🌟New
#CHI2024
Paper.
Thrilled to announce our new research on cross-cultural differences in public perceptions of AI conversational agents between the US and China (w/ Zihan Liu,
@AnfanChen
,
@renwenzhang
, Yi-Chieh Lee). More in this thread 🧵
Preprint Alert🤖 Our latest review
@Nature_NPJ
shows that AI chatbots significantly reduce distress, especially in multimodal, generative AI chatbots & in clinical and elderly groups. Key to success? Quality human-AI interactions and engaging content:
Our latest JCMC paper explores the emotional implications of AI companionship by analyzing 35,000+ posts from r/replika. We uncover seven types of human-AI interactions and their links to user emotions, revealing the bittersweet "paradox of emotional connection with AI". 🤖❤️
🚀Preprint alert! Excited to share our latest paper (with
@renwenzhang
) accepted by
@ica_jcmc
. By analyzed 35k+ screenshots and posts from Replika users, we uncovered 7 types of daily human-AI interactions and the paradox of emotional connection with AI.
Computational Summer Institute is back in Singapore! If you are passionate about computational methods in social sciences, don’t miss out - apply by March 31! Stay tuned for speaker updates. More info 👉
#SICSSSingapore2024
I passed my dissertation defense today! A huge thanks to my fantastic committee (Madhu Reddy,
@DavidCMohr
,
@nataliebazarova
,
@dagwood
), and to my mentors, colleagues, and friends for their intellectual and emotional support throughout and well beyond graduate school!
New paper with my PhD student
@zichengstephy
and our CNM graduate Dr. Yuren Qin is out in JCMC! In this paper, we examine factors predicting toxicity and prosocial behavior in online gaming environments, highlighting the pivotal role of interdependence among game players. (1/2)
We’re happy to announce a new
#jcmc
publication titled “Toxicity and Prosocial Behaviors in Massively Multiplayer Online Games: The Role of Mutual Dependence, Power, and Passion” by
@zichengstephy
,
@renwenzhang
, and Yuren Qin! Read the article here: [] 1/4
Kicking off the 2nd Summer Institute in Computational Social Science (SICSS) Singapore with a fascinating keynote from Dr. Subhayan Mukerjee
@wrahool
@CNMnus
Computational Summer Institute is back in Singapore! If you are passionate about computational methods in social sciences, don’t miss out - apply by March 31! Stay tuned for speaker updates. More info 👉
#SICSSSingapore2024
🚨SICSS Singapore returns for the third year running! 🚨
Grad students, postdocs, and early career faculty - if you're interested in computational methods in the social sciences, apply by March 31!
Speakers to be confirmed soon - more details 👇
#SICSSSingapore
Day 5:
@YingdanL_kk
captivated us with her fascinating research on political visual analysis and multimodal computational research. Her talk highlights the immense potential of large multimodal models (LMM) for social science research.
Do you know who's in the house? 🏡🔥 It's
@renwenzhang
! 📖🤩 Don’t miss the opportunity to catch her presentations at
#ICA24
and walk away with greater insights than before. 🌟
#CNM
#NUSResearch
New pub alert🚩 We surveyed 1131 Wuhan residents, including essential workers, about their social media use during Covid. We found that passive social media is related to both healthy and unhealthy social comparisons, depending on social context (ie, quarantine status).
Excited to share our new paper in JCMC. In this article,
@iPhiloSophia1
and I examined Facebook users’ privacy management strategies in times of stress. We found that people disclosed more on Facebook despite privacy concerns when they were highly stressed.
“Privacy Management and Self-Disclosure on Social Network Sites: The Moderating Effects of Stress and Gender” by Renwen Zhang and Jiawei Sophia Fu, Journal of Computer-Mediated Communication,
We concluded with a compelling talk by
@alvinyxz
on computational storytelling. He highlighted various projects, each starting with data, method, and question. Key message is that great computational research finds a perfect balance between these three elements. 🌟
#SICSSSingapore
Day 6: An enlightening session with
@Diyi_Yang
on leveraging LLMs in computational social science! Key insights:
1️⃣ Prioritize open-source LLMs for classification.
2️⃣ Conduct thorough cost-benefit analyses.
3️⃣ Stay vigilant about biases and potential harms.
#SICSSSingapore
Day 1 -
@ChaoYu24
delivered an informative and engaging workshop on text sentiment analysis. The participants had a great learning experience!
Then in the afternoon, we have
@ChaoYu24
(
@cnmnus
) talking about "Exploring Subtle Motivations and Unintended Consequences Using Text Sentiment Analysis"
#SICSS
-Singapore Day
#2
Thrilled to have Dr. Cindy Shen
@cuihua
enlighten us with an insightful talk about the journey and future trajectory of computational communication🚀 Future directions include multimodal data analysis and powerful synergy of big data + survey/experiment.
Happy to share our
#chi2021
paper co-authored with
@liltove
@DavidCMohr
, Madhu Reddy & Melina Pann: We talk about how mental health technologies can better integrate system-initiated suggestions and user-initiated customization without overburdening users.
Currently in the Mental Health (A) session! I'm in on the author side of this platform and wow it's a whole other experience. But yay
@renwenzhang
presenting our paper on designing for emotional well-being!
#CHI2021
#1
Many survey platforms (eg, Qualtrics) offer fraud detection functions, and they are handy! ALWAYS activate IP address screening and bot detection features like CAPTCHA to block repeat offenders and bots (3/9).
We’ll be hosting SICSS again in Singapore this summer! If you’re interested in computational methods. don’t miss out this exciting opportunity. Registration link below:
Super excited to announce that SICSS Singapore is running again this year! Grad students, postdocs, and early career faculty - if you're interested in computational methods in the social sciences, apply by March 6, 2023!
More details here:
A big shout out to my wonderful students and collaborators
@zichengstephy
@lexaaylexaay
Xinyuan, and Chris! Catch the incredible
@zichengstephy
for her two presentations about Tiktok health comm & problematic gaming today and tomorrow!
In our
#CHI2021
paper co-authored with
@nataliebazarova
and Madhu Reddy, we examined the factors that influence online disclosures of COVID-related distress, highlighting the pivotal roles of perceived affordances and relational closeness to the audience.
Thought the cleaning was over? Think again. Odd patterns in regression diagnostics and low reliability scores in established scales raised red flags 🚩. Turns out, data wrangling isn't just about skills, it's about vigilance.
#ResearchEthics
(9/9)
Excited to share a preprint of our
#DIS2022
paper on designing for digital wellbeing and technology non-use. Super proud of my student
@zichengstephy
!
How can people build a healthy, intentional relationship with technology? CNM PhD candidate Zicheng Zhu, along with
@lexaaylexaay
and
@renwenzhang
, explored this question in the game context in their new paper accepted by
#DIS2022
(1/4)
In our new study
@ijhcs_journal
, we found that AI face-editing techs encode gender biases, reinforcing a "programmed masculinity" & "gender panopticon" that monitors and restricts gender expressions.
#DesignEquity
#InclusiveTech
#2
Eliminate near-duplicate responses. While not carbon copies, these entries show high similarity—often 80-90% identical across certain sections—suggesting duplication (4/9).
In the afternoon,
@he_tianyu
shared her latest research on AI-assisted theory building and experimental studies. Exploring AI’s role in advancing scientific knowledge and innovation is crucial and promising.
#SICSSSingapore
Day 2 -
@chrischaosu2021
talked about his cool projects on fact-checking co-citation networks and platform policy evolution. He shared a super useful tool for archiving and analyzing platform policy changes:
#4
Scrutinize responses from identical latitude and longitude, as they may signal bots or fraud. But sometimes legit users have location services off. Here's a trick: check the timestamps. Loads of responses from one spot in a flash are more like a bot bash (6/9).
Thanks Mora for making such an amazing podcast episode! So honored to be part of it and connect to the Spanish audience - this really motivates me to learn Spanish😆
#SICSSSingapore
Day 6: An enlightening session with
@Diyi_Yang
on leveraging LLMs in computational social science! Key insights:
1️⃣ Prioritize open-source LLMs for classification.
2️⃣ Conduct thorough cost-benefit analyses.
3️⃣ Stay vigilant about biases and potential harms.
In the afternoon session, Dr. Prasanta Bhattacharya captivated us with an engaging lecture on social network analysis, bringing the concepts, methods, and applications to life with vivid examples. It was an inspiring session that left us with plenty of food for thought!
#SICSSSingapore
Day 3: Data annotation is essential for NLP/CSS research.
@jiaxin_pei
shared valuable strategies for effective data annotation, including quality control measures, and demonstrated his customizable text annotation tool POTATO:
#SICSSSingapore
Day 4:
@kous2v
blew us away with his research on measuring well-being using social media and multimodal data! He tackled big questions like causal inference, ethical issues, and mitigating harms in computational methods.
#5
Scrutinize the content, especially open-ended answers. Are the answers varied and detailed, or do they seem generic and repetitive? That's your sign (7/9).
#SICSSSingapore
Day 2 -
@chrischaosu2021
talked about his cool projects on fact-checking co-citation networks and platform policy evolution. He shared a super useful tool for archiving and analyzing platform policy changes:
Happy to announce our CHI 2020 Workshop, Technology Ecosystems: Rethinking Resources for Mental Health, is now open for submissions. Papers are due Feb 11th. Come join us in Hawai'i!
#CHI2020
2 years ago, I posted an online survey that got 981 responses in just 4 hrs. *spoiler alert* almost all of them were from bots 🤖
Sound familiar? Check out our case study paper on fraud prevention techniques, coming to
#chi2024
!
pre-print:
The deadline extended to June 7! Come and join our
#ICA24
preconference on computational message science on June 19 at QUT💡Check out our fantastic speakers lineup of speakers, and register here:
#SICSSSingapore
Day 3: Data annotation is essential for NLP/CSS research.
@jiaxin_pei
shared valuable strategies for effective data annotation, including quality control measures, and demonstrated his customizable text annotation tool POTATO:
Our opening keynote on Monday (June 3) is by
@noshir
(
@NorthwesternU
) who will talk about "Digital Exhaust from the Web to Leverage Network Insights in the Algorithmically Infused Workplace"
CFP📢 We're editing a special issue for Frontiers in Computer Science/Psychology on the topic of "Responsible AI in Healthcare". The submission deadline is 15 August 2022. See more details here: DM me if you're interested or have any questions.
📢 I’m co-editing a special issue for Frontier on the topic of
#HumanCenteredAI
and
#Healthcare
. Let me know if you are interested in submitting or reviewing papers.
Responsible AI in Healthcare: Opportunities, Challenges, and Best Practices
This is the most efficient and painless submission experience I've ever had, taking less than 6 months from the initial submission to final acceptance. Big shout out to JCMC editors
@nicole_ellison
@cuihua
, editorial assistant Brenna Davidson and the anonymous reviewers!
Check out this amazing work by
@yixwang
@ndiakopoulos
on journalists' sourcing needs and practices using user-generated content and the design of a UGC sourcing tool for journalists
#CHI2021
Really enjoyed writing up the article about online communities and mental health with
@chengchengQu
for this new issue of
@XRDS_ACM
on "Is Computing Toxic? Thank
@Orson_Xu
for the invitation and the amazing XRDS team! Online copy available here:
Guest editor
@Orson_Xu
along with lead editor
@xjasminelu
knocked it out of the park with this Fall's XRDS issue on "Is Computing Toxic?". Online copy available here:
So grateful to my fellow co-organizers
@renwenzhang
@jacklqiu
for helping make it happen, and
@chris_bail
@msalganik
for the vision! It's truly incredible how SICSS has spread around the world since its inception back in 2017. We'll be back next year! 🖖
Kicking off the first ever SICSS-Singapore with
@sgonzalezbailon
introducing the field of Computational Social Science and her personal trajectory in the field to a full room of participants (for an 8:30 am session)!
The stellar
@joejps84
@sheavze
and Prasanta Bhattacharya from
@ASTARsg
held an engaging afternoon workshop on using large language models in social science. Check out their latest paper on using LLMs for psychological measurement from text: .
We build on and extend the theory of situational privacy and self-disclosure by
@MasurPhil
, the disclosure decision-making framework by
@naz_andalibi
, personal social media ecosystems by
@DrMichaelAnn
et al., and other inspiring work on online disclosure and privacy
Can't believe we are already more than half way through the
@oiioxford
#oiisdp
2019. As ever we have an outstanding group of visiting PhD scholars conducting cutting-edge research on societal uses of digital technologies.
#internetschool
Meanwhile, conversational AI showed no significant improvement in psychological well-being. More research is needed to explore the long-term effects of AI chatbots on mental health, the underlying mechanisms of such effects, and the safe integration of LLMs in mental health care.
@naz_andalibi
@NorthwesternU
@pitch_lab
Thank you for facilitating the great discussion of stigmatized disclosure on social media. My students really enjoyed reading your paper and talking with you!
@JaeJaeykim2
@J_Chris_Li
@YingdanL_kk
Thanks for your interest! Since the speaker's talk includes work-in-progress, we prefer to keep the slides internal. Feel free to check out Dr. Lu's publications, which were covered in her talk:
My first solo-author CHI paper, "A Place to Play: The (Dis)Abled Embodied Experience for Autistic Children in Online Spaces" is a follow-up to my work with the
@AutCraft
community. Looking forward to sharing with
#chi2019
this spring!
@mkrawczyk23
Thank you for sharing! Such a timely and crucial paper - wish I'd come across it sooner. Fully on board with the call for early-stage fraud prevention and a systematic approach to safeguard data integrity. Could you share your Participant Validation Guide not linked here?
@naz_andalibi
Thank you Naz! This totally resonates with me. It's been really difficult to read these reactions and also some news reports, while my family and friends are looking for ways to donate and support our fellow citizens who are suffering.
As
#CHI2021
is ongoing, time to introduce three CHI papers published in last three years - all from the
#onlinedeliberation
#Singapore
project.
1. a platform
2. some interface nudges
3. a simple Q&A nudge
🌟New
#CHI2024
Paper.
Thrilled to announce our new research on cross-cultural differences in public perceptions of AI conversational agents between the US and China (w/ Zihan Liu,
@AnfanChen
,
@renwenzhang
, Yi-Chieh Lee). More in this thread 🧵
@jdfoote
Thank you, Jeremy, for raising this issue. It's indeed tough to balance false positives and false negatives. I've tried to eliminate obvious bots (like multiple responses with similar patterns) while keeping unique responses, but it's not perfect. Any suggestions are welcome.