Drew Kelly
About
Advancing AI and data science research, education, policy, and practice to improve the human condition
Advancing AI and data science research, education, policy, and practice to improve the human condition
The Stanford Institute for Human-Centered AI (HAI) is an interdisciplinary institute that advances AI and data science research, education, policy, and practice. Launched in 2019, HAI draws on faculty and students from all seven of Stanford’s schools — business, education, engineering, humanities and sciences, law, medicine, and sustainability — along with partners across industry, government, and civil society, all united by the idea that AI should enable human flourishing.
The most important questions in AI do not respect disciplinary boundaries. They require experts across fields to work together at a scale that individual labs or departments cannot achieve alone. Stanford HAI brings together the interdisciplinary people, expertise, and shared capabilities that make this work possible.
AI is a civilization-changing technology. It will change how we work, communicate, learn, create, and make decisions, reshaping individual lives, institutions, and society itself. Stanford HAI was created in anticipation of this moment, and our work is organized around a conviction that shapes everything we do: AI should be guided by its human impact (H), designed to augment rather than replace people (A), and inspired by human intelligence (I).
We pursue that mission through three pillars — accelerating discovery, transforming education for human flourishing, and understanding and shaping AI for the good of society — each grounded in a cross-cutting commitment to openness. Open science, open source, and open class are what universities uniquely contribute to a rapidly evolving field. We publish in open forums, we release our models, tools, and datasets, and we make our teaching accessible worldwide.
AI is a general purpose technology that can be used for good and for ill. Our vision for the future is led by a commitment to human-centered uses of AI and data science and to ensuring that the benefits are broadly shared. In support of these goals, our research falls into three pillars: AI and data science for accelerating discovery, transforming education, and understanding and shaping AI for societal good.
Open scholarship is academia’s core advantage. HAI begins with openness — in the research we conduct and in the models, datasets, software, benchmarks, and educational resources we share. It takes three forms across everything we do.
Open science. Shared datasets, benchmarks, and open publications — including the work of establishing a science of AI measurement, with methods and standards the whole field can use.
Open source. Software, data, models, and infrastructure that serve the broader ecosystem. We build genuinely open-source, domain-specific models, supported by open-source specialists on staff.
Open class. K–12, university, policy, and executive education that translates Stanford expertise worldwide.
None of this works in isolation, so we build collaborative networks with aligned organizations globally. Shaping AI for humanity cannot mean shaping it for one region or one worldview.
The Stanford HAI leadership share a vision of AI and data science serving the collective needs of humanity. It is this goal that guides the work of HAI.
Human-Centered AI is about augmenting rather than replacing humans so that these technologies can enhance what makes us fundamentally human — our empathy, creativity, judgment, and wisdom.
AI is a civilizational technology. HAI serves as a lighthouse for human-centered AI — expanding the frontiers of knowledge and reimagining how we learn, use and govern this technology, with human dignity and agency as our North Star.
2019:
HAI officially launches; our leaders propose a national AI research resource (NAIRR) to drive American innovation in AI, and we launch a fellowship program for faculty.
2020:
We launch the Stanford Digital Economy Lab with economist Erik Brynjoffson, issue our first Hoffman Yee Grants for asteroid shot ideas in AI, and partner with Stanford CS and the McCoy Center for Ethics in Society to launch Stanford's Embedded EthiCS program.
2021:
We pilot the first Ethics & Society Review for awarding grants, launch the Center for Research on Foundation Models with Percy Liang at the helm, and create a new graduate fellowship program for Stanford students.
2022:
We launch our policy boot camp to education regulators, debut our new student affinity groups, and create a new Industrial Affiliate Program to work with industry on human-centered AI.
Christine Baker
2023:
We partner with Wu Tsai Center to fund research in AI and neuroscience, develop our Tech Ethics & Policy fellowships, launch the RAISE-Health initiative with Stanford Medicine, create an AI training series for federal employees, and team with the Asia Foundation for an education program in Asia. HAI leaders Fei-Fei Li and Rob Reich also meet with Pres. Biden to discuss American innovation in AI.
2024:
HAI cohosts the Joint California Summit on Generative AI with California Governor's Office of Business and Economic Development (GO-Biz), the California Government Operations Agency, and the UC Berkeley College of Computing, Data Science, and Society
2025:
NVIDIA’s Yejin Choi joined as the Dieter Schwarz Foundation HAI Professor, Professor of Computer Science, and Stanford HAI Senior Fellow. The Stanford AI Lab (SAIL) joined forces with HAI. And Fei-Fei Li won the Queen Elizabeth Prize for Engineering.
2026:
Stanford HAI and the Stanford Data Science initiative merge into a single institute under the Stanford HAI name, uniting more than 400 scholars, $60 million in cumulative grant funding, and the Marlowe computing cluster. James Landay continues as Denning Director; Fei-Fei Li becomes Special Advisor on AI to President Levin and joins President Emeritus John Hennessy as co-chair of the HAI advisory council.
Learn about the Stanford Institute for Human-Centered AI (HAI) and our mission through the voices of our leaders.
The 2024-25 year marked a significant period of growth as we continued to advance the mission of human-centered AI. Government leaders, industry executives, educators, and civil society organizations have come to HAI for guidance on both technical insight and a deeper understanding of what AI means for their communities and institutions.
AI can accelerate scholarship in every discipline — and domain-specific scholarship is itself key for advancing AI. We work to catalyze that virtuous cycle.
Better AI makes better scholarship possible. Less obviously, the reverse is also true: the hardest problems in, for example, astronomy, medicine, economics, and the humanities expose the limits of current models and point toward what AI needs to become. HAI is built to close that loop.
Our work here has two dimensions. AI-driven discovery advances every discipline, where domain-specific scholarship can itself become a powerful vector for advancing AI — from decoding the physical universe from data, to understanding how ideas, language, and cultures evolve, to designing environments and therapies for human health. Our research on Foundations of AI pursues new architectures, algorithms, and evaluation approaches that go beyond simply adding silicon to scale models, including the search for shared principles underlying natural and artificial intelligence.
As AI reshapes work and society, what should people learn, and how should they learn it? We are building the rigorous evidence base to answer these questions.
Most conversations about AI in education begin and end with how AI transforms schooling. We start from a broader goal: supporting human flourishing in the age of AI. That reframing changes the research agenda.
We ask what knowledge and skills learners of all ages actually need as AI reshapes work, civic life, and the professions. We also focus on how to learn: developing AI-enabled learning systems that genuinely support students, teachers, communities, and institutions.
Answering these questions requires evidence gathered over time. We aim to build longitudinal research infrastructure — with Stanford’s own courses serving as a testbed — to track learner outcomes through their time at Stanford and well beyond graduation. This work extends from K–12 through university and lifelong learning.
AI is scaling at a breakneck pace while public trust in its governance remains unsettled. HAI offers a rigorous, empirically grounded alternative to the extremes dominating public conversation.
To shape AI for positive societal impact, you need to understand both how the systems work and how they transform incentives, institutions, and human lives. That requires technical and humanistic expertise working in dialogue — not going along parallel tracks. That integration is what Stanford HAI was built to provide.
Our research fuses technical AI expertise, design, and humanistic inquiry to study AI systems as they actually behave in the world and to develop new methods and tools to design them for the better. Then we translate that scholarship into clear, trusted guidance that leaders can depend upon to make decisions as AI continues to evolve.
Get the latest news, advances in research, policy work, and education program updates from HAI in your inbox weekly.
Sign Up For Latest News
Open to Stanford community members! We are hosting an interactive orientation featuring faculty insights, networking, and opportunities to explore AI and data science research, programs, and resources across Stanford.
In a new series called Prompt Response, Stanford’s Surya Ganguli, Diyi Yang and Rob Reich examined emergent agent behavior, recursive self-improvement, independent evaluation and whether a kill switch can make advanced AI safer.
As AI moves beyond language into systems that can perceive, understand, and act in the physical world, a new frontier is emerging: world models—AI systems that build and maintain working representations of real environments to predict how they change in response to action.
