---
title: README
emoji: 🤝
colorFrom: indigo
colorTo: pink
sdk: static
pinned: false
short_description: Bidirectional Human-AI Alignment Lab @ NYU Shanghai & NYU
---

<p align="center">
  <img src="https://hua-shen.org/assets/images/lab_logo_full.png" alt="BiAlign Lab" width="420"/>
</p>

# 👏 Welcome to BiAlign Lab!

**BiAlign Lab** at **NYU Shanghai & NYU** studies how humans and AI interact with and adapt to each other, with a focus on **bidirectional human-AI alignment**.

Our work connects **Machine Learning**, **Multi-Modal AI (NLP, Speech, Vision)**, **Human-Computer Interaction**, and **Social Computing**. We aim to build AI systems that are intelligent and that also respond to human values and societal needs, toward more **trustworthy and responsible AI**.

## 🔬 Research Themes
- **Bidirectional Human-AI Alignment**: aligning AI to humans *and* helping humans adapt to AI
- **Value Alignment**: the value–action gap, value trade-offs, and value steering in LLMs
- **AI Safety**: interpretability, steerability, and the gap between what agents claim and what they do
- **Human-Centered AI**: AI for caregiving, mental health, education, and older adults

## 🌐 Community
- 📄 [NeurIPS 2025 Position Paper: Bidirectional Human-AI Alignment](https://arxiv.org/pdf/2406.09264)
- 🧑‍🤝‍🧑 [BiAlign Workshop @ ICLR & CHI 2025](https://bialign-workshop.github.io/)
- 🎓 [NeurIPS 2025 Tutorial](https://nips.cc/virtual/2025/tutorial/109592)
- 📚 [BiAlign Course @ NYU Shanghai](https://hua-shen.org/src/course_bialign.html)

## 📦 What You'll Find Here
Datasets, models, and demos released with BiAlign Lab papers. Each repo includes a card that describes the data or model, its intended use, its license, and how to cite it.

## 🔗 Links
- 🏠 Lab website: [hua-shen.org/src/lab.html](https://hua-shen.org/src/lab.html)
- 👩‍🏫 PI: [Hua Shen](https://hua-shen.org/)

## 🙌 Join Us
We welcome PhD, Master's, and undergraduate researchers, as well as visitors and collaborators. See [Join Us](https://hua-shen.org/src/lab.html) for details.