Sherry Tongshuang Wu

Hello World!
I'm Sherry Tongshuang Wu
(吴彤霜)!

Assistant Professor

School of CS, Carnegie Mellon University (CMU SCS)

Human Computer Interaction Institute (HCII)

Language Technololgy Institute (LTI)

I am trained (by my amazing PhD advisors Jeffrey Heer and Dan Weld at the University of Washington) to be an HCI+NLP researcher. I study how humans (AI experts, lay users, domain experts) interact with (debug, audit, collaborate) AI systems.

Most recently, I work on:

Build practical AI systems, by mapping general-purpose AIs to the right specific use cases.

Click & jump to some recent papers that represent my research interests and style:

Watch: Real-world AI Evaluation

We askWhat can general-purpose models do?

We doReplicate diverse human-subject experiments with general-purpose models.

Inspect: Task-specific AI Test & Distill.

We askHow to map general purpose models to special purposes?

We doPerform task-specific model testing and distillation. Design measurements that quantify user-specific net gains.

Enrich: Human-AI Task Delegation

We askHow to map models to the right purposes so they are helpful?

We doFind optimal sub-tasks for AIs and humans; Explore venues where AI mistakes can be features; Instruct humans to recover from AI errors.

If you are interested in exploring relevant topics with me at CMU, I will be looking for undergraduate, master or PhD students! PLEASE read this FAQ to find out our open projects and best ways to contact us.

Research Highlights

Beyond Relevance: Evaluate and Improve Retrievers on Perspective Awareness

Xinran Zhao, Tong Chen, Sihao Chen, Hongming Zhang, Tongshuang Wu

CoLM 2024: Conference on Language Modeling

What Is Wrong with My Model? Identifying Systematic Problems with Semantic Data Slicing

Chenyang Yang, Yining Hong, Grace A. Lewis, Tongshuang Wu, Christian Kästner

ASE 2024: the 39th IEEE/ACM International Conference on Automated Software Engineering

LLMs as Workers in Human-Computational Algorithms? Replicating Crowdsourcing Pipelines with LLMs

Tongshuang Wu, Haiyi Zhu, Maya Albayrak, Alexis Axon, Amanda Bertsch, Wenxing Deng, Ziqi Ding, Bill Guo, Sireesh Gururaja, Tzu-Sheng Kuo, Jenny T Liang, Ryan Liu, Ihita Mandal, Jeremiah Milbauer, Xiaolin Ni, Namrata Padmanabhan, Subhashini Ramkumar, Alexis Sudjianto, Jordan Taylor, Ying-Jui Tseng, Patricia Vaidos, Zhijin Wu, Wei Wu, Chenyang Yang

ArXiv 2024: ArXiv

Promp2Model: Generating Deployable Models from Natural Language Instructions

Vijay Viswanathan, Chenyang Zhao, Amanda Bertsch, Tongshuang Wu, Graham Neubig

EMNLP Demo Track 2023: The 2023 Conference on Empirical Methods in Natural Language Processing

Beyond Testers' Biases: Guiding Model Testing with Knowledge Bases using LLMs

Chenyang Yang, Rishabh Rustogi, Rachel Brower-Sinning, Grace Lewis, Christian Kaestner, Tongshuang Wu

EMNLP Findings 2023: The 2023 Conference on Empirical Methods in Natural Language Processing

Is AI the Better Programming Partner? Human-Human Pair Programming vs. Human-AI pAIr Programming

Qianou Christina Ma, Tongshuang Wu, Kenneth Koedinger

AIED2023 Empowering Education with LLMs 2023: AIED2023 Empowering Education with LLMs - the Next-Gen Interface and Content Generation

What You Say = What You Want? Teaching Humans to Articulate Requirements for LLMs

Qianou Ma, Weirui Peng, Hua Shen, Kenneth Koedinger, Tongshuang Wu

ArXiv 2024: ArXiv 2409.08775

How to Teach Programming in the AI Era? Using LLMs as a Teachable Agent for Debugging Best Paper

Qiaomu Ma, Hua Shen, Kenneth Koedinger, Tongshuang Wu

AIED 2024: The 25th International Conference on Artificial Intelligence in Education

Sherry @ CMU

Research Highlights