The Godmother of AI

As the Sequoia Capital Professor in Computer Science at Stanford University and co-director of the Stanford Institute for Human-Centered Artificial Intelligence (HAI) — an interdisciplinary research hub that engages the social sciences and humanities — she’s spent more than 25 years doing just that while driving breakthroughs in cognitively inspired AI, computer vision, deep learning and AI systems for healthcare delivery. She is an elected member of the National Academy of Engineering, the National Academy of Medicine and the American Academy of Arts and Sciences. In 2023, she was the recipient of the Intel Innovation Lifetime Achievement Award.

On Feb. 24, Li will be honored with the Woodrow Wilson Award, given annually on Alumni Day to celebrate a Princeton undergraduate alumna or alumnus whose career embodies the call to duty in Wilson’s 1896 speech, “Princeton in the Nation’s Service.” The 2024 award honors a career dedicated to advancing AI research while also advocating for the ethical and responsible use of AI technologies in improving the human condition. 

Mistrust of artificial intelligence is on the rise thanks to ChatGPT, the popular chatbot created by OpenAI, and other generative AI tools triggering concerns over machines taking jobs, spreading disinformation and threatening human extinction. “I know AI is a topic that stirs up people’s emotions,” Li said. “As long as we maintain that humanity in us, I have confidence we will navigate our world through the AI era and make AI a force for good.” 

During her Princeton talk and in subsequent media interviews, Li has drawn a distinct separation between AI being used to support and enhance humans and AI being used to replace them. On the augmenting human side lie deep learning systems that can help scientists track contamination in natural waterways and recognize aggressive cancer cells. On the replacing side, for example, are AI loan review processes that perpetuate biases and discrimination in credit decisions. 

At this point, artificial intelligence is still limited in what it can do. That’s why Li believes now is the ideal time to set up guardrails for how AI is developed, deployed and governed. “We have to frame AI as a tool because there’s so much talk about AI as this nebulous terminator,” Li said in McCosh. Like other tools that have served human interests like steam engines, electricity and computers, “AI can be used for the public good, especially if we ensure that human agency always remains valued,” she said. 

Li has both laid the foundation for the AI revolution and pushed to ensure the revolution is carried out responsibly. It all goes back to something she learned in college, Li told her McCosh audience. As she came to grips with the profound impact AI would have on the world, she thought of Princeton’s informal motto about service: “It made me realize that I have a responsibility to make a good future out of this, not only as a technologist but as a Princetonian to be in the nation’s service and the service of humanity.” 

Teaching Computers to See 

Fei-Fei Li grew up in Chengdu, China, before emigrating with her parents to New Jersey in 1992 when she was 15 years old. Her parents encouraged her love for science and technology while she attended Parsippany High School, and she supplemented the family income with a part-time job in a Chinese restaurant. All of this Li explores in her recently published autobiography “The Worlds I See: Curiosity, Exploration, and Discovery at the Dawn of AI,” which intertwines her coming-of-age story as a Chinese emigrant with the steady maturation of AI. The book was recently selected as the Princeton Pre-read for the class of 2028.

Li overcame plenty of obstacles — such as her parents not speaking English — in adapting to life in the U.S. and pursuing a career in science. In her book, she acknowledges all the help she had along the way, like her high school math teacher, Bob Sabella, who taught her advanced calculus during lunch hours. “I’m optimistic about humanity because my own journey has shown me human kindness, compassion, integrity and generosity,” she said during her November presentation. 

She eventually earned a scholarship to attend Princeton where she studied computer science, engineering and physics: “What I loved about physics was its audacity to ask the most profound and daring questions about our universe.” Questions such as: How did time begin? What is the universe’s unifying force? And what does it mean to see? “When I do my research, I try to be fearless about asking the most audacious questions,” she said. 

While pursuing her Ph.D. in computer science and neuroscience at Caltech, Li became interested in how the brain processes visual information. Her doctoral work involved developing algorithms that recognize objects within images, laying the groundwork for her later research in computer vision, which has led to advancements in areas such as self-driving cars, facial recognition systems and medical imaging technologies. 

Li’s visual algorithm research shifted into overdrive when, as an assistant professor at Princeton in 2007, she began working on the project that would become ImageNet, the groundbreaking online image database that helped kickstart the deep learning revolution. “ImageNet was both a training dataset as well as a benchmark to a fundamental problem in machine learning: recognizing objects in the world,” Li said. “For that time, it was a huge data effort in which we open-sourced the dataset — 15 million images across 22,000 categories, after all the curation — and open-sourced a competition [the ImageNet Challenge] for global researchers that eventually led to deep learning.” 

The ImageNet team would grow to include both Stanford and Princeton computer scientists, including Kai Li, the Paul M. Wythes ’55 and Marcia R. Wythes Professor in Computer Science, Jia Deng *12, associate professor of computer science, and Olga Russakovsky, associate professor of computer science, all now Princeton faculty. 

AI’s Big Bang moment came when neural network algorithms running on powerful computer processors, originally designed to render video game graphics, used ImageNet’s massive dataset to train computers to make sense of the visual world. “Before ImageNet, machine learning was focused on very small amounts of data,” Li said. “Image classification changed the way machine learning was done.” 

AI for Everyone 

“I wrote this memoir because I wanted to communicate what I think AI truly is, which is a science that is curious and whimsical instead of something scary,” Li said during her November campus visit. “I also wrote it to tell the human story of AI through a voice that’s lacking in today’s AI world.” 

She noted that AI still lacks diversity, with prominent figures being predominantly male. This lack of diversity can hinder scientific progress by limiting perspectives and approaches to problem-solving. 

To help address AI’s lack of diversity, Li teamed up with Russakovsky, her former student, in 2017 to establish AI4ALL, a nonprofit organization initially aimed at promoting gender equality in the field of artificial intelligence. “When I pitched the idea to Fei-Fei, she was so excited she almost fell out of her chair,” Russakovsky said. “The first year was just a two-week day camp for 24 high school girls and it was just amazing.” 

Since then, AI4All has formed partnerships with universities across the country to provide college-level AI programs for students from historically underrepresented groups and free curriculum resources that educators can use to increase access to AI education in their communities. “Those are the groups we’ve identified where we can have the most impact,” Russakovsky said. “We have a great curriculum and plenty of industry partners helping out.” 

Building on her mission to ensure that diverse voices are heard and represented in scientific research, Li also established HAI in 2019 alongside John Etchemendy, former provost at Stanford University and the Patrick Suppes Family Professor at Stanford’s School of Humanities and Science. “HAI’s goal is to put the well-being of individuals and society at the center of designing, developing and deploying of AI technology,” Li said. 

One of the founding missions of HAI is to create a bridge between Silicon Valley and the policymakers in federal and state governments. In 2016, Li was appointed by President Barack Obama to serve on the White House’s National Science and Technology Council’s Select Committee on STEM Education. She has also served as special adviser to the Secretary General of the United Nations and a member of the National Artificial Intelligence Research Resource Task Force for the White House and the National Science Foundation. She continues working with policymakers nationally and locally to ensure the responsible use of technologies, including testifying before Congress in 2018 regarding AI’s power and responsibility in the application of advanced technologies, saying, “There’s nothing artificial about AI. It’s inspired by people, created by people, and most importantly, it has an impact on people.” 

Godmother of AI 

Li still has plenty of work ahead of her, especially when it comes to ensuring that universities and other institutions continue to play a significant role in AI research. Unfortunately, large-scale machine learning, the breakthrough technology that gives computers the ability to recognize images and respond to natural language, requires more computing power than most universities have access to. “I feel strongly about the utter imbalance of public-sector resources in AI innovation versus industry,” she said from the stage at McCosh 50. “Not a single university in America has the resources to train a ChatGPT model.” 

Li has called for federal investment in public-sector AI research. AI is too powerful, she argues, to relinquish its development to deep-pocketed tech companies. “The public sector, especially universities, is like a garden of 1,000 flowers and AI is such fertile ground for so many different ideas,” Li said. “The work of universities in producing knowledge-driven research is more critical than ever.” 

Somewhere along the path of her pioneering career, experts in her field began to refer to her as the godmother of AI as a tip of the hat to her groundbreaking contributions. It is a title she didn’t immediately embrace. “I would never call myself that,” she told CBS News last year. “I don’t know how to balance my personal discomfort with the fact that, throughout history, men are always called godfathers of something.” 

“Fei-Fei deserves the godmother of AI title and I’m glad she’s now accepting it,” Russakovsky said. “I’m also glad that AI has a recognized godmother because there have been lots of women who have substantially contributed to this field. Fei-Fei was willing to engage with a lot of the difficult issues in AI before it became popular.” 

Li continues to engage with difficult issues in her own uniquely thoughtful way. “I do see myself with the responsibility of continuing to push forward a human-centered AI framework,” she said in her speech in McCosh 50. “Because humans need to enter the era of intelligent machines able to develop and deploy this technology in a responsible way that works for everyone.” 

Alumni awards and accompanying lectures will be presented in Richardson Auditorium during Alumni Day ceremonies that also will recognize student winners of the Jacobus Fellowship and Pyne Prize. The Alumni Day program will also feature the annual Alumni Association luncheon at Jadwin Gymnasium and the Service of Remembrance at the Princeton Chapel, honoring alumni, students, and members of the Princeton University faculty and staff whose deaths were recorded by the University in 2023.