BrightUp Institute

Science for human well-being in the age of AI

BrightUp Institute is the research division of BrightUp Kids. We connect rigorous science, emerging technology, and community experience to advance human development, mental health, and well-being.

Our work has a particular commitment to children, young people, families, and underserved communities. We pursue research that produces useful evidence and resources for the public, not only for a single product or organization.

Partner with us Our research programs

How we work

Three commitments run through everything we take on.

Public-benefit researchQuestions chosen for what the answer is worth to the public, not to any one product or organization.
Community participationFamilies, practitioners, and communities help shape the questions and take part in the work.
Transparent methodsWe work in the open, so that methods, code, and findings can be checked and reused.

Two research programs

One program studies people; the other studies the technology now entering their lives. Healthcare is a major application area in both. The two are organizing structures, not silos: a study of how AI responds to a teenager in distress belongs to both at once.

Human Development & Mental Health

Central question

What helps people develop, adapt, and thrive, and what prevents them from receiving effective support?

We study how people develop and thrive, how family and social environments shape well-being, and how access to effective support can be improved. Our work brings empirical methods together with the experiences of families, communities, and practitioners.

Where we begin

Developmental trajectories
Family and social environments
Mental-health needs
Access to effective support

Children and young people come first; caregivers and adult populations are included where the science calls for it. Not every project needs an AI component.

AI & Human Well-Being

Central question

How can AI advance human well-being, and how do we establish whether it actually does?

We develop and evaluate AI methods that can improve people’s lives. Our research examines model capabilities, human-AI interaction, healthcare and mental-health applications, and the conditions under which AI is useful, reliable, and beneficial.

Where we begin

Model and multimodal methods
Human-AI interaction
Health-information quality and mental-health support
Open research infrastructure: benchmarks, datasets, software

We both build and evaluate: developing useful methods, and testing whether they help in the real world. “Responsible” is a standard we hold the work to, not a limit on what we study.

Research that others can build on

We aim to share findings through scientific publications, technical reports, benchmarks, research software, and accessible public resources. Data sharing follows applicable permissions, privacy protections, and ethical requirements.

Publications and technical reports
Peer-reviewed papers and reports written to be checked.
Benchmarks and evaluation datasets
Validated with experts, reproducible, and open for comparing any system.
Research software
Code others can run, inspect, and extend.
Public resources
Plain-language summaries and materials for families and practitioners.

Where we are starting

Two initiatives in development. Neither is complete or externally funded.

In development

A Video Benchmark for Multimodal AI in Child Development

How well do multimodal AI systems understand a child’s development from video, and how sound is the guidance they give parents? The benchmark asks both. Its first release focuses on early signs associated with autism, in English, and pairs labels designed for validation by clinical and developmental experts with reproducible evaluation code, so any system can be compared on the same footing.

In development

Family Research Advisory Panel

A standing group of families and people with relevant lived experience who help set the Institute’s research priorities, review study designs, and take part in the work, with training and support along the way. It gives BrightUp Kids’ community a real seat in the research rather than a survey at the end.

People

The Institute is led by its Founding Scientific Director and draws on BrightUp Kids’ leadership and board.

Marcus Xue, Ph.D.
Founding Scientific Director
Leads the Institute: research strategy, AI methods, publications, and research partnerships. Previously a postdoctoral researcher at Stanford University and an AI data science manager in technology; founder of AI ECD Inc., an AI startup in the San Francisco Bay Area.
Prof. Yaobo Shi
Director of Empirical Research
Leads empirical design and methods. Two decades of research on how public goods and public policies shape people’s lives over the long run.
Jing Zhang
President, BrightUp Kids
Community partnerships, program operations, and participant engagement.

Partner with us

We welcome collaborations with researchers, healthcare organizations, community groups, and funders. Partnerships begin with a clear research question, defined responsibilities, and a commitment to credible findings.

A few sentences is enough. It helps if you cover:

Who you are and the organization you work with.
The question you want to answer, or the work you want to contribute to.
What a good outcome would look like for the people involved.

Email the Institute

info@brightupkids.org

A person from our team replies, usually within a few days.

BrightUp Institute is the research division of BrightUp Kids, a registered 501(c)(3) nonprofit organization, and is not a separate legal entity. BrightUp Kids is the legal party to every agreement, application, and publication. The Institute conducts research and does not provide clinical, counselling, or emergency services. If you or someone you know needs support now, contact a healthcare professional or local emergency services.