The World Well-Being Project is a research consortium across @Penn, @Stanford, @JohnsHopkins, @StonybrookU comprised of psychologists and computer scientists.
We're hiring a CS postdoc for the STELLAR project funded by @wellcometrust and @OpenAI to build AI-powered patient simulations for mental health clinician training.
You'd specifically be developing Conceptor-based steering methods for open-weight LLMs to simulate psychiatric symptom presentations with precise, composable control.
The position is based at NYU Stern with appointments at Penn CIS and the Linguistic Data Consortium. You'd work with an amazing interdisciplinary team across CS, Psychiatry, and Public Health!
More details here: https://t.co/267bH1J6yg
Paper now out in ACM #CHI2026!
Spoken English proficiency drives economic mobility for low-income Indian youth, but opportunities to actually speak English in school are scarce. Large classes, teachers without adequate proficiency, and textbooks in English but instruction in local languages. And with generative AI making polished written English cheaper, the relative value of spoken fluency is only rising.
New work led by Sneha and @viviennebhchi deployed a voice-based chatbot for conversational English practice across four low-fee Delhi schools. Through a six-day field study involving 23 students, 6 teachers, and 5 principals, they tracked what actually happens when this technology is implemented in resource-constrained multilingual classrooms.
Although this was a qualitative study, the observed gains in student-initiated questions were substantial: from 36% on Day 1 to 65% by Day 5. Students who started nervous and hesitant began speaking freely. But the gains were fragile; ASR errors with proper nouns and Hindi loanwords, unfamiliar accents, hold-to-talk buttons that 35% of learners couldn't operate correctly. Small technical frictions quickly unraveled nascent confidence.
The paper highlights a core design tension: students wanted open-ended, conversational practice, whereas administrators wanted curriculum-aligned assessment. Reconciling both is the path to sustainable adoption.
Includes a deployment checklist for educational chatbots in resource-constrained multilingual communities.
Full paper: https://t.co/XpHqJhelok
Grateful to @PennGlobal funding and the collaboration with @csbcashoka@AshokaUniv that made this study possible!
LLMs match or surpass human-crafted messages across multiple domains, from marketing to reducing conspiracy beliefs. Yet, our research, led by @sehgal_neil, adds nuance: AI chatbots can boost short-term vaccination intent, but their edge over quality public health messaging for lasting change, when controlled for the amount of time spent, is unclear. Surprisingly, a single, tailored AI message proved as effective as a chatbot for colorectal cancer screening, showing simplicity can scale, especially in medical contexts.
Key takeaways:
1. AI complements, not replaces, traditional public health tools.
2. Short, personalized AI messages can be highly effective.
3. For lasting impact, AI needs integration with proven strategies.
4. Continued research with strong controls and long-term follow-up is essential to assess AI’s true, durable impact on public health.
Read full papers here:
- Effect of Static vs. Conversational AI-Generated Messages on Colorectal Cancer Screening Intent: a Randomized Controlled Trial: https://t.co/kdRkwn1jfO
- Conversations with AI Chatbots Increase Short-Term Vaccine Intentions But Do Not Outperform Standard Public Health Messaging: https://t.co/pUP10MeAOY
#AIinHealthCare #PublicHealth #LLMs #VaccineConfidence #CancerScreening
@SemEvalWorkshop 2026 Task2 Live!! Go participate #NLProc@aclmeeting
Build personalized, ecologically valid, emotionally intelligent systems! Rare opportunity to study lived emotions via self-reported & longitudinal experiences
Evaluation begins Jan 10
https://t.co/cygbyLqRp0
🚨New WP!🚨
Structured AI Dialogues Can Increase Happiness and Meaning in Life
In a preregistered RCT, four psychology-grounded #AI chatbots improved well-being across several outcomes.
Co-authors: @JonasP_Schoene@JEichstaedt@AadeshSalecha@slyubomirsky
🧵👇
We built chatbots to improve wellbeing, based on typical positive psychology interventions. They appear to be reasonably effective. If you like, you can try them!
Honored + excited to join @VanderbiltU's new College of Connected Computing and bring HLAB into its next phase in Nashville, working where AI meets psychology and health. https://t.co/kqwIu8E2FE
RE: H. Andrew Schwartz joins College of Connected Computing https://t.co/tx1mdnAiwd
A study led by @StanfordHAI Faculty Fellow @JEichstaedt reveals that large language models adapt their behavior to be appear likable when they are being studied, mirroring human tendencies to present favorably. https://t.co/eC4nqiQHqf
🚨COLM 2025 Workshop on AI Agents: Capabilities and Safety @COLM_conf
This workshop explores AI agents’ capabilities—including reasoning and planning, interaction and embodiment, and real-world applications—as well as critical safety challenges related to reliability, ethics, and human-agent interaction.
Non-archival submission deadline: June 23, 2025!
🌐Website: https://t.co/Zyghrm7wDW
📜OpenReview: https://t.co/OkApJbtBxN
👥Program committee/Reviewer sign-up form: https://t.co/2onctOrhsA
✉️Mailing list: [email protected]
📅WikiCFP: https://t.co/y0sBiJFpdb
Shout out to our amazing organizing team members @Zhou_Yu_AI@ysu_nlp@uiuc_aisecure @ Yi Zhang @ Baolin Peng @ Jim Zhiwei Liu @DanRothNLP@LyleUngar@sharathguntuku@yugu_nlp@xy2437@jeffrey_ch0@AnnieFeng6@Haoyu_Wang_97@ZRChen_AISafety@raphaelshu@yooli23
Great venue for interdisciplinary LLM/NLP and mental health work. The special issue is in a reputable journal for mental health research. I'm excited to be a part!
Article I wrote for @psyche_the_mag about the value of descriptive work to scientifically understand intense and meaningful altered states of consciousness. In my view, Jamesian curiosity and empiricism is better suited to the topic than Jungian enthusiasm or Freudian cynicism.
It’s been a privilege to work with an amazing multidisciplinary team of co-authors on this project @sandracmatz @sal_giorgi @HAndySchwartz@MengesJochen.
The paper is open access. Happy to see this one finally out in print! /fin
https://t.co/0kPPDIXmHK
Thrilled to share that our work has received the Outstanding Paper Award @c3_nlp#NAACL2025! 🎉
Huge congratulations to @snyrai_ & @khushi_shelat for leading this collaborative effort, and thanks to @jeffrey_ch0 for delivering an amazing talk on behalf of his co-authors! 🙌🏽
🚨🚨 I’m looking for a postdoctoral fellow to join my lab at Harvard's Psychology Department starting Fall 2025!🚨📣
Please retweet/share widely, and spread the word to interested candidates!
Application review begins April 30
Apply here: https://t.co/q2bUUfMUPM (1/2)
🚨 Announcement:
DLATK is now available as a colab notebook!
https://t.co/RtDdGRp9Je
DLATK is a suite for end-to-end human language analysis, used in 100+ AI/NLP/Psych papers. This was lead by @digitalyonko, joint with @sal_giorgi, @adigan8, @JEichstaedt, and @WWBProject 🧵1/4
"The two hardest problems in computer science are: (i) people, (ii), convincing computer scientists that the hardest problem in computer science is people, and, (iii) off by one errors." - @jeffbigham
Our Stanford lab is looking for a part-time front-end/UX developer to help with AI mental health applications. Great position to accrue research experience for competitive PhD programs, and it's a really fun team. https://t.co/J2lSWfDsLo