free web templates

Photo credit: Aurora Brachman

serinac@berkeley.edu
Office: Berkeley Way West 8060
Pronouns: she/her
[CV] [Google Scholar] [X]

Serina Chang

I am an Assistant Professor at UC Berkeley, jointly appointed in EECS and Computational Precision Health and part of the Berkeley AI Research (BAIR) Lab. If you're interested in working with me, please see Getting Involved below for details!

My research falls at the intersection of AI and human behavior, including modeling human behaviors with AI, improving and evaluating human-AI interaction, and developing AI tools for societal decision-making, with a focus on public health. My work is recognized by the KDD Dissertation Award, KDD Best Paper Award, Forbes 30 under 30, Google Research Scholar Award, EECS Rising Stars, and Rising Stars in Data Science, and has been featured by over 650 news outlets, including The New York Times and The Washington Post. Previously, I completed my PhD in CS at Stanford University, advised by Jure Leskovec and Johan Ugander, and I was a postdoc at Microsoft Research in the Computational Social Science group.

Recent News

Teaching

Research

Some research directions I'm currently excited about include (but are not limited to):

1) Building AI for human interaction. How do humans interact with generative AI in the real world and how do we incorporate human interaction into AI evaluation and development? How do we adapt AI to understand individuals' diverse values and needs and to respond in high-stakes contexts?
 Examples: converting static benchmarks into human-AI interactions and simulating users (ACL'25), probing LLMs' knowledge of human opinions (ICLR'26), building politically neutral AI (ongoing), assessing AI impacts on human health (ongoing).

2) Simulating behaviors with generative AI. How do we build more realistic, diverse, and scalable simulations of human behavior? How do we validate AI simulations and safely integrate them into real-world pipelines?
 Examples: generating social networks with LLMs (ICWSM'25), predicting public opinions with LLM fine-tuning (ACL'25), graph-based approaches to simulation (preprint), allocating human samples for fine-tuning vs statistical corrections (preprint).

3) Inferring real-world behaviors from novel data sources. How do we infer hard-to-observe behaviors from novel data sources? How do we leverage aggregated or unlabeled data, balance privacy needs, and validate our inferences?
 Examples: inferring dynamic mobility networks and modeling COVID-19 spread (Nature'21, ICML'24), inferring demographic-specific mobility trajectories with aggregate supervision (under review), estimating vaccine uptake and hesitancy from anonymized search logs (Nature Comm'24), measuring immigration attitudes from political speeches (PNAS'22).

4) Supporting public health and policy with AI. How can AI and behavioral modeling help to guide more effective and equitable decision-making? How do we translate scientific advances into real-world decision-support tools?
 Examples: mobility-based reopening dashboard for Virginia Dept of Health (KDD'21), vaccine site recommendations (IAAI'22), spillover effects of pandemic policies (AAAI'23), AI for infectious disease modeling (Nature'25), collaboration with United Nations Development Programme (ongoing).

Getting Involved

Prospective postdocs (position available!)
 I am recruiting a postdoc working in AI & society to start in summer 2026. The position is fully funded for two years, so a 2-year postdoc is preferred, but 1-year postdocs will also be considered. The official posting will be released soon.

Prospective PhD students
 I can advise PhD students at UC Berkeley in EECS or Computational Precision Health (CPH). I'm generally interested in students who are skilled in ML and data science, care deeply about real-world impact and interdisciplinary work, have strong critical thinking skills, and are clear communicators. A strong research fit is also important, such as interests in AI, human behavior, networks/graphs, public health, computational social science, or any of the research directions described above.

Fall 2026 applications: Given the volume of applications, including my name makes it much likelier that your application will be flagged to me. For EECS or CPH, you can mention me by choosing my name when entering your faculty preferences on your application; you can also mention my name in your Statement of Purpose. For EECS, I will also be paying special attention to applications that select "AI-H" as their primary area. For CPH, I will be paying special attention to applications that select "Community and Public Health" under Health Research Topics.

UC Berkeley undergrads and master's students
 If you are interested in a research assistant position, please fill out this form (requires UC Berkeley log-in). Applications are open on a rolling basis, but we're likelier to have open positions if you apply before the start of the semester. I am typically unable to work with undergraduate or master's students who are not UC Berkeley students.

Selected Papers

Please see my CV for a full list of papers.

Rethinking LLM human simulation: when a graph is what you need
Joseph Suh, Suhong Moon, and Serina Chang
Under review
[paper] [code]


Learning demographic-conditioned mobility trajectories with aggregate supervision
Jessie Li, Zhiqing Hong, Toru Shirakawa, and Serina Chang
Under review


What do large language models know about opinions?
Erfan Jahanparast, Zhiqing Hong, and Serina Chang
ICLR 2026


ChatBench: from static benchmarks to human-AI evaluation
Serina Chang, Ashton Anderson, and Jake Hofman
ACL 2025 (main)
[paper] [data] [code]


Language model fine-tuning on scaled survey data for predicting distributions of public opinions
Joseph Suh*, Erfan Jahanparast*, Suhong Moon*, Minwoo Kang*, and Serina Chang
ACL 2025 (main)
[paper] [data] [code]


LLMs generate structurally realistic social networks but overestimate political homophily
Serina Chang*, Alicja Chaszczewicz*, Emma Wang, Maya Josifovska, Emma Pierson, and Jure Leskovec
ICWSM 2025
Presented at IC2S2 2024 as Plenary Talk (2.8% submissions)
[paper] [code]


Artificial intelligence for modelling infectious disease epidemics
Moritz U. G. Kraemer*, Joseph L.-H. Tsui*, Serina Chang*, Spyros Lytras, Mark P. Khurana, Samantha Vanderslott, Sumali Bajaj, Neil Scheidwasser, Jacob Liam Curran-Sebastian, ..., and Samir Bhatt*
Nature 2025
[paper]


Measuring vaccination coverage and concerns of vaccine holdouts from web search logs
Serina Chang, Adam Fourney, and Eric Horvitz
Nature Communications 2024
Also presented at KDD 2023 Workshop on Epidemiology Meets Data Mining and Knowledge Discovery (oral) and KDD 2023 Workshop on Data Science for Social Good (oral)
[paper] [data & code]


Inferring dynamic networks from marginals with iterative proportional fitting
Serina Chang*, Frederic Koehler*, Zhaonan Qu*, Jure Leskovec, and Johan Ugander
ICML 2024
Also presented at Learning on Graphs 2023 (extended abstract)
[paper] [code]


Computational analysis of 140 years of US political speeches reveals more positive but increasingly polarized framing of immigration
Dallas Card, Serina Chang, Chris Becker, Julia Mendelsohn, Rob Voigt, Leah Boustan, Ran Abramitzky, and Dan Jurafsky
PNAS 2022
Article in Stanford HAI News by Edmund L. Andrews
[paper] [code]


Supporting COVID-19 policy response with large-scale mobility-based modeling
Serina Chang, Mandy L. Wilson, Bryan Lewis, Zakaria Mehrab, Komal K. Dudakiya, Emma Pierson, Pang Wei Koh, Jaline Gerardin, Beth Redbird, David Grusky, Madhav Marathe, and Jure Leskovec
KDD 2021, Applied Data Science Track - Best Paper Award
[paper] [code] [blog post]


Mobility network models of COVID-19 explain inequities and inform reopening
Serina Chang*, Emma Pierson*, Pang Wei Koh*, Jaline Gerardin, Beth Redbird, David Grusky, and Jure Leskovec
Nature 2021
Commentary in Nature News and Views by Kevin Ma and Dr. Marc Lipsitch
Interactive article in The New York Times by Yaryna Serkez
Selected media coverage: The New York Times, The Washington Post, Bloomberg, CNN, Wired, MIT Technology Review, and Stanford Press
Also presented at Networks 2021 (oral), NeurIPS 2020 Workshop on Machine Learning for Health, NeurIPS 2020 COVID-19 Symposium, and OECD-ODISSEI Webinar on Open Data Infrastructure
[paper] [code] [talk] [website]


© Serina Chang 2026