iiinova — where your digital self keeps evolving

Brihi Joshi @brihi.bsky.social · 29/04/2025

📖 So fun to collaborate with amazing folks at USC (Xiang Ren, @swabhs.bsky.social ) and Apple (Rik Koncel-Kedziorski, Tim Paek) on this work! 🤝💡📣 Excited to finally share this—let us know what you think! 🙌🏽✨🎉 More details in our draft here: arxiv.org/abs/2504.17993 (8/n)

arxiv.org

Improving LLM Personas via Rationalization with Psychological Scaffolds

Language models prompted with a user description or persona can predict a user's preferences and opinions, but existing approaches to building personas -- based solely on a user's demographic attribut...

021

Brihi Joshi @brihi.bsky.social · 29/04/2025

🌎 PB&J improvements do not hinge on any particular group—we see gains across education, race, income, and gender. 🌈📊🤝 (7/n)

100

Brihi Joshi @brihi.bsky.social · 29/04/2025

💡 It’s not just about having more reasoning tokens—it’s about what those tokens convey. 🧐🔍✨ We found that *human-written rationales* improved personas the most, but LLM-generated ones came close, showing that good scaffolds are very useful for personas. 🧑‍💻📈🔬 (6/n)

100

Brihi Joshi @brihi.bsky.social · 29/04/2025

📊 We tested PB&J on two tasks: predicting opinions and movie ratings. PB&J consistently outperformed methods based on demographics and/or judgments. Using scaffolds boosted performance even more! 🎥🗳️🏆 (5/n)

100

Brihi Joshi @brihi.bsky.social · 29/04/2025

🔧 We propose psychological scaffolds—frameworks based on life experiences, personality traits, or core beliefs—to further guide the structure of LLM rationales. 🛠️🧬🎨 (4/n)

100

Brihi Joshi @brihi.bsky.social · 29/04/2025

🧠 We introduce PB&J (Psychology of Behaviour and Judgments) that draws from folk psychology—the way people naturally explain each other’s actions. PB&J adds LLM-generated rationales to personas, helping models better represent user behavior. 📚🧩💬 (3/n)

110

Brihi Joshi @brihi.bsky.social · 29/04/2025

🤖 LLMs can mimic human behavior with personas, but most current approaches focus only on demographics or past responses as user context, often overlooking the reasons why users think or act a certain way. 🧐💭🔄 (2/n)

100

Brihi Joshi @brihi.bsky.social · 29/04/2025

Reasoning about the "why" behind user behavior can improve LLM personas! ✨🧠📈 📝Excited to share our new work: Improving LLM Personas via Rationalization with Psychological Scaffolds 🔗 arxiv.org/abs/2504.17993 🧵 (1/n)

1134

Brihi Joshi @brihi.bsky.social · 21/11/2024

omggg it’s amazing 😭 combined together gave me queen vibes!!

010

Brihi Joshi @brihi.bsky.social · 19/11/2024

🙋🏻‍♀️🙋🏻‍♀️

110

Brihi Joshi @brihi.bsky.social · 17/11/2024

omg this can be 3D printed at the makerspace!!

020

Brihi Joshi @brihi.bsky.social · 17/11/2024

These days, I’ve been obsessed with reading about how people use LLMs and this blog by Nicholas Carlini has been my TOP read since August - nicholas.carlini.com/writing/2024... Not a lot of us who work with LLMs “use” them, and this is such a great starter to all of that!

nicholas.carlini.com

How I Use "AI"

I don't think that AI models (by which I mean: large language models) are over-hyped. In this post I will list 50 ways I've used them.

0100

Reposted by Brihi Joshi

🐰𝓑𝓾𝓷𝓷𝔂🐰 @xnightbloombunny.bsky.social · 15/11/2024

Netflix needs to stop asking if I’m still watching and start asking if I moved the laundry to the dryer

683408114972

Reposted by Brihi Joshi

Swabha @swabhs.bsky.social · 14/11/2024

#EMNLP2024 has been a treat so far! Of course it doesn't hurt to win an outstanding paper award with my incredible PhD students Jaspreet Ranjit and @brihi.bsky.social and our wonderful collaborators at USC 😜 Paper: dill-lab.github.io/oath-frames/

0171