Image: Fei-Fei Li at AI for Good 2017 by ITU Pictures (via Flickr), licensed under CC BY 2.0.
Dr. Fei-Fei Li is a pioneer in computer vision who helped establish image recognition as a big part of modern AI. She’s the “Godmother of AI” due to her work on image-based systems. Li became the co-founder and chairperson of AI4ALL, a nonprofit that improves accessibility in AI education. Her work influenced conversations about how AI should be developed and taught. While working as a professor at Stanford University, she said that teaching machines to see is one part of AI’s progress as we learn to develop, use, and guide the technology responsibly.
Based on her background, we can see why Li views AI as a human and technical problem. She became well-known for her help with making computer vision a serious field of AI. Before earning her undergraduate degree in physics from Princeton University, she completed her PhD at Caltech. With her training in physics, she modeled complex systems for vision and learning. Throughout her career, Li worked at institutions like the University of Illinois Urbana-Champaign and Stanford, where she had a big role in AI research and helped advance human-centered AI efforts. At all these institutions, she pushed AI forward while questioning how it may affect people.
Her most well-known achievement, ImageNet, helped reshape computer vision and became one of the main reasons the field rapidly progressed in the 2010s. Li and collaborators developed the large visual dataset and benchmark so computers could be taught to identify objects from images at scale. It has millions of images within thousands of categories like chairs, dogs, tools, and vehicles. Before ImageNet, teams would use different datasets and testing, which made results comparison across studies more difficult. Thanks to ImageNet, researchers used a shared testbed and a massive labeled dataset that accelerated progress. It helped unlock deep-learning breakthroughs via the large-scale data for more powerful model training.
(Image Credit: Igor Omilaev/Unsplash)
Fei-Fei Li focused on the real-world effects of AI once ImageNet helped launch the modern era of computer vision. She spoke about AI’s risks, fairness, transparency, and the impact of automated decision-making. While AI became more powerful, Li was presented with a new challenge: ensuring the technology serves people’s needs rather than performance optimization. With that goal in mind, she co-founded Stanford’s Institute for Human-Centered AI (HAI), an institute that focuses on developing AI systems to respect human values and remain accountable to those it affects.
Teachers might see this change as an important lesson and a powerful teaching model, as it pairs technical learning with human insight. For example, in the classroom, a computer vision lesson doesn’t need to start with complex algorithms. Instead, it may start with easy-to-understand examples, like identifying objects in images. Students see how images are grouped, how labels are chosen, and how minor decisions define an AI system’s behavior.
ImageNet also makes it easier to understand AI concepts, making it relevant in education. Machine learning can be hard for students to grasp. Providing them with a set of images and instructing them to create categories can help them learn more about it. After labeling them, the students compare decisions with an AI model. When they see differences in classifications, students start discussing how humans and machines interpret the data differently.
(Image Credit: Markus Winkler/Unsplash)
Her work shows that it’s important to understand bias in AI. Because ML models are trained from existing data, they can reproduce the limitations and patterns found in that data. Stanford HAI’s CRAFT curricula help teachers teach students how to examine bias. Students then ask questions like, " Who is represented in a dataset? Who is missing in it? How do different perspectives influence the way information is labeled?” Having conversations like these can show how AI systems are shaped by human decisions and the data they learn from.
Li’s contributions are significant as AI is part of everyday life. It’s used for search engines, image and text generation, making predictions, and more. And although students frequently interact with AI tools, they may not know how these systems generate outputs or recommendations. Teaching datasets, training, and classification basics allows students to understand the technology and use the system more responsibly. Additionally, those skills can help them ask important questions about AI systems they will use in the future.
She has emphasized that AI is not an all-knowing machine. Stanford’s AI + Education program supports educators in building AI literacy and critical thinking about AI systems. While teaching machine learning or computer vision, it helps to start with the people, choices, and consequences behind the technology. This allows students to better understand how AI systems are created and how they affect the world. In 2025, Fei-Fei Li received the Queen Elizabeth Prize (QEPrize) for Engineering along with other AI pioneers.
Overall, Li has shown that AI research advancements and AI education have deep connections. Teachers rely on Li’s approach to show that understanding AI isn’t about explaining how it works, but helping students understand its impact.
Have a story tip? Message me here at element14.