"AI Godmother" Fei-Fei Li's Latest Interview: I Didn't Expect AI to Become So Popular, the Next Frontier Is Spatial Intelligence

Fei-Fei Li is a professor at Stanford University, known as the “Godmother of AI.” She gained fame for releasing the ImageNet visual database in 2006, which propelled the development of AI. Recently, Fei-Fei Li appeared on a podcast to discuss AI, spatial intelligence, and superintelligence.

• She believes AI is a double-edged sword, with enormous potential but also possible risks, and emphasizes that humans should remain in control of AI development.
• Fei-Fei Li hopes AI technology will become more democratized, so that everyone has the ability to influence and use it responsibly, rather than it being monopolized by a few large tech companies.
• She proposed that the next frontier of AI is “spatial intelligence”, meaning AI’s ability to understand, perceive, reason, and interact with the three-dimensional world, which complements visual intelligence.
• She pointed out that AI’s profound impact on education, employment, and society needs to be handled with caution, and believes that individuals, enterprises, and society all have a responsibility to address the job reshaping brought by technology.
• Regarding the potential risks of superintelligence, she believes that the problem does not lie with the machines themselves, but with how humans manage and govern the technology, as well as international responsibility and regulation.
• She maintains a pragmatic attitude towards energy consumption and environmental impact, believing that renewable energy and innovation in energy policy are indispensable parts of AI development.
• She emphasizes the importance of returning to traditional values and educational concepts: cultivating children’s curiosity, critical thinking, and sense of responsibility, rather than relying on AI to be lazy or abusing technology.

The following is the translated transcript of Fei-Fei Li’s conversation. For brevity and clarity, the content has been edited.

Q: May I start with this extraordinary period in your industry? It’s been three years since ChatGPT was released to the public. Since then, there have been new tools, new applications, and massive investments. How would you describe your feelings at this moment?

Fei-Fei Li: AI is not new to me. I’ve been in this field for 25 years, working with it every day since the start of my career. However, the depth and breadth of its current impact still astonish me, and sometimes it feels unreal. This is a civilization-level technology. I am one of the scientists who helped bring it into being, but I never expected it to become so immense.

Q: When did everything start to change? Was it because of the speed of technological development, or because the world finally woke up and shone a spotlight on you?

Fei-Fei Li: I think it’s a combination of both. But when I say it’s a civilization-level technology, it’s not because we’re getting attention or because it’s so powerful, but because it affects so many people. Everyone’s life, work, well-being, and future will be touched by AI in some way.

Q: Including negative impacts?

Fei-Fei Li: Technology has always been a double-edged sword, right? Since the dawn of human civilization, the tools we’ve created—technology—have generally been used for good, but they can also be deliberately misused or have unintended negative consequences.

Q: You mentioned “powerful.” Today, the power of this technology is in the hands of a very few companies, most of which are American. What do you think about this?

Fei-Fei Li: You’re right. Large tech companies have the greatest impact on society through their products. Personally, I hope this technology can become more democratized. Whoever builds or possesses it should use it responsibly. I also believe everyone should have the ability to influence this technology.

Q: You’re both a scholar and a tech company CEO. Your company has only been around for a little over a year, and reportedly has a valuation of 1.1 billions.

Fei-Fei Li: Yes! (laughs) I’m the co-founder and CEO of World Labs. We are pioneering the next frontier of AI—spatial intelligence. Nowadays, everyone talks about large language models, so this term isn’t heard as much. But I believe spatial intelligence is just as critical as language intelligence, and they complement each other.

Q: I know your first love was physics.

Fei-Fei Li: Yes.

Q: Among the physicists you admire most, what was it in their lives or work that made you go beyond physics itself?

Fei-Fei Li: I grew up in a not-so-well-known city in China, in a very ordinary family, so my world was quite small. My childhood was simple and closed off; I was an only child. Physics was the opposite—it was vast, bold, and boundless in imagination. Looking up at the stars, I could ponder the origins of the universe; observing a snowflake, I could delve into the molecular structure of matter; thinking about time, magnetic fields, atomic nuclei... Physics took my imagination to places I could never reach in this life. What still fascinates me is that physics taught me to dare to ask the boldest, most audacious questions about the physical world, the universe, and our origins.

Q: And your own “audacious question,” I suppose, is “What is intelligence?”

Fei-Fei Li: Yes. From Newton to Maxwell, from Schrödinger to Einstein—my most admired physicists—I appreciate the “audacious questions” each of them posed.

I also wanted to find my own. Midway through university, my questions shifted from the physical world to intelligence: What is intelligence? How does it arise? Most fascinating of all, how can we create intelligent machines? That became my pursuit, my North Star.

Q: That’s a quantum leap—from machines that can only compute to machines that can continuously learn and keep learning.

Fei-Fei Li: I like your use of “quantum leap” as a physics pun.

Q: I noticed you read very widely. The breakthrough came when you started paying attention to research by psychologists and linguists related to your field.

Fei-Fei Li: That’s the charm of doing science at the frontier. It’s all new; no one knows how to do it. Naturally, I studied the human brain and mind for inspiration. When I was trying to crack the puzzle of visual intelligence early on, one inspiration was studying how humans organize the semantic space of vision. There are thousands, millions of objects in the world—how are they organized? Alphabetically? By size? By color?

Q: Did you study this because you had to understand how the brain organizes information before you could teach computers?

Fei-Fei Li: You could put it that way. I came across a linguistic achievement called WordNet, which organizes semantic concepts using a specific taxonomy—not visual, just semantic or lexical.

Q: Can you give an example?

Fei-Fei Li: In a dictionary, “apple” and “appliance” are close together, but in real life, apples and pears are closer—they’re both fruits, while appliances are a completely different category. I had a flash of insight: maybe the organization of visual concepts is like this, with apples and pears more closely related than apples and washing machines. More importantly, the scale. When you see the sheer number of objects language can describe, you realize that as intelligent beings, we experience the world through massive amounts of data—we need to give machines this ability too.

Q: It’s worth noting that at that time—the early 2000s—the concept of “big data” didn’t exist.

Fei-Fei Li: The term hadn’t even appeared yet. The scientific datasets we used then were very small.

Q: How small?

Fei-Fei Li: In terms of images, the datasets used by our generation of graduate students usually had only four to six, at most twenty, object categories. Three years later, we created ImageNet: 22,000 object categories, 15 million labeled images.

Q: ImageNet was a major breakthrough and the reason you’re called the “Godmother of AI.” I wonder, what enabled you to make connections others didn’t see? You came to the US at 15, English was your second language—did that contribute to the kind of inspiration you mentioned?

Fei-Fei Li: I don’t know. Human creativity is still a mystery. Some say AI can do everything—I disagree. There’s still so much we don’t know about the human brain. I can only speculate that it’s a combination of my interests and experiences. I dare to ask “crazy” questions in science and am not afraid to think outside the box. My sensitivity to the connection between language and vision may well come from my own experience learning a foreign language.

Q: Coming to the US as a teenager is hard enough, but you also had to overcome a language barrier. What was that like?

Fei-Fei Li: It was hard. (laughs) I came to the US at 15, to Parsippany, New Jersey. None of my family spoke good English. I learned quickly, but my parents struggled a lot. We were financially strapped; my parents worked as cashiers, and I worked in a Chinese restaurant. When I went to college, my mother’s health was poor, so we decided to open a dry-cleaning shop to make ends meet.

Q: Did you personally participate?

Fei-Fei Li: I joke that I was the CEO. From age 18 to mid-grad school, I ran it for seven years.

Q: Even when you weren’t there, you managed it remotely?

Fei-Fei Li: Yes. I spoke the best English, so I handled all customer calls, bills, and quality checks.

Q: What did that experience teach you?

Fei-Fei Li: Resilience. You need resilience to do scientific research, because the path of science is nonlinear—no one has ready answers. As an immigrant, you also have to learn to be resilient.

Q: Did your parents have high expectations for you?

Fei-Fei Li: To be fair, they didn’t push me much—they weren’t “tiger parents.” They just wanted to survive. My mother is an intellectual at heart and loves to read, but immigrant life was hard and her health was poor, so she had no time to manage me. As a teenager, I had no choice: succeed or fail, with high risk, so I was self-driven. I was always curious, and science became my outlet and my anchor. I wasn’t into clubbing, just science.

Q: Was there a teacher who was important to you?

Fei-Fei Li: I was good at math and became friends with my math teacher, Bob Sabella, because we both loved science fiction. At first, I read Chinese sci-fi, then English. He saw my thirst for knowledge and created opportunities for me. After I skipped a grade in math and had no more classes to take, he tutored me one-on-one during lunch, with no extra pay. That was a teacher’s love, and also a sense of responsibility. He became a benefactor in my life.

Q: Is he still alive?

Fei-Fei Li: He passed away when I was an assistant professor at Stanford, but his wife and children became my “family” in New Jersey.

Q: Did they help you integrate into American society?

Fei-Fei Li: Of course. They showed me what a typical middle-class family looked like and were my window into American society.

Q: Do you think you could have had the same career in China?

Fei-Fei Li: Life is full of chance, so I can’t assume. But what doesn’t change is curiosity and the pursuit of my North Star. So I believe I would have devoted myself to AI no matter what.

Q: Do you still have connections with China?

Fei-Fei Li: That’s my root. I’m lucky to have developed in the US. Now I live in Stanford, San Francisco, Silicon Valley—it’s very international here; AI as a discipline is also global. I’m more like a global citizen.

Q: China’s AI patents and papers have surged, and Deepseek drew attention earlier this year. What do you think of US-China AI competition?

Fei-Fei Li: China is undoubtedly an AI powerhouse. Currently, the consensus is that the US and China are the two leading countries in AI. Many regions around the world aspire to have a place in AI.

Q: What do you mean by the next frontier, “spatial intelligence”?

Fei-Fei Li: Spatial intelligence is AI’s ability to understand, perceive, reason, and interact with the world—it’s the continuation of visual intelligence. The first half of my career was about solving the problem of “seeing,” which is passive information reception. But in evolution, intelligence is inseparable from action: we need to see because we move, and we move better because we see. How do we build this connection? We need to understand 3D space, how objects move, how I reach out to grab a cup—the core of all this is spatial intelligence.

Q: The Marble demo on your website is a virtual world. Is it a tool for training AI?

Fei-Fei Li: Let me clarify the definition. Marble is a cutting-edge model that can generate 3D worlds from simple prompts. For example, “give me a modern kitchen,” or give it a photo of a kitchen, and it can generate a 3D world. Humans are naturally able to create 3D worlds, and I hope AI can too. Designers can use it for ideation, game developers can quickly get 3D scenes, robots can use it for simulation training, and it can also be used for AR/VR education.

Q: I can imagine Afghan girls attending class in a virtual classroom.

Fei-Fei Li: Yes. Or how to help an 8-year-old understand “cells”? In the future, we could build a world inside a cell, let students walk in, see the nucleus, enzymes, membranes—the possibilities are endless.

Q: The industry faces many real-world issues. First, will AI destroy a large number of jobs?

Fei-Fei Li: Technology will change the landscape of labor. AI is so profound that it will inevitably have a profound impact on employment. Salesforce CEO Marc Benioff said 50% of the company’s customer service jobs have already been handed over to AI. This is really happening. Every time humanity invents more advanced technology—steam engines, electricity, computers, cars—we experience pain, but also job reshaping. It’s not enough to just discuss the increase or decrease in job numbers; we need a more nuanced response: individuals must keep learning, and enterprises and society also have responsibilities.

Q: Second, Nobel laureate Geoffrey Hinton believes there’s a 10%–20% chance AI could lead to human extinction. What do you think?

Fei-Fei Li: First, Geoff is my friend of 25 years—I respect him, but we often debate. On “replacing humanity,” I respect but disagree. It’s not impossible, but if humanity really faces a crisis, it will be because of our own mistakes, not the machines. He raises practical issues: when superintelligence is born, how do we prevent it from taking over? We have no precedent. My question is: Why would humanity as a whole allow this to happen? Where is our collective responsibility, governance, and regulation?

Q: Can we put a cap on superintelligence?

Fei-Fei Li: I believe we can ensure that technology is developed and used responsibly at the international level.

Q: At the government level, is it through treaties? Or just companies agreeing to act in a certain way?

Fei-Fei Li: This field is still in its infancy; we don’t have international treaties or that level of global consensus yet. But I think we already have global awareness. What I want to say is, we shouldn’t over-worry about the negative impact of AI. This technology is very powerful. It may bring some negative consequences, but it also brings many applications that benefit humanity. We need to look at the big picture.

Q: I know you often communicate with politicians—in the US, UK, France, and elsewhere. What’s the most common or frustrating question they ask you?

Fei-Fei Li: I wouldn’t use the word “frustrating”—I’d say “worrying,” because I think our public discussion about AI needs to go beyond the question, ‘What do we do when machine overlords appear?’ Another question parents around the world often ask: AI is here, what should my child do? Should they study computer science? Will there be jobs in the future?

Q: How do you answer?

Fei-Fei Li: AI is a powerful technology, and I’m a mother too. The most important thing is to raise children as “humans”: give them agency, dignity, curiosity, and eternal values like honesty, diligence, creativity, and critical thinking. Don’t just worry about majors; understand your child’s interests and personality, and guide them accordingly. Anxiety solves nothing.

Q: I have another industry question about the huge influx of capital into companies like yours. Will there be a bubble, like the internet bubble, causing some companies to be overvalued?

Fei-Fei Li: First, my company is still a startup. When we talk about huge sums of money, we often focus on large tech companies. AI is still an emerging technology, with many areas needing development. This science is very complex, and breakthroughs require massive resources. That’s why it’s crucial to provide resources for this research. On the other hand, there’s the market: can we see returns? I believe the application prospects for AI are very broad—software engineering, creativity, healthcare, education, financial services—the market will continue to expand. AI can be an assistant or collaborative tool, helping people improve well-being and productivity, meeting many needs. I firmly believe this part of the market will keep growing.

Q: But what price will be paid in terms of electricity, energy, and climate? A well-known AI entrepreneur, Jerry Kaplan, said that because huge data centers consume massive amounts of energy, we may be heading for a new ecological disaster.

Fei-Fei Li: That’s a very interesting question. To train large models, we need more and more electricity or energy, but no one says these data centers have to use fossil fuels. Innovation in the energy sector will be a key part of this.

Q: The amount of electricity they need is so huge; it’s hard to meet the demand with renewables alone.

Fei-Fei Li: That’s true at present. Countries building these large data centers also need to review their energy policies and industrial structures. This gives us an opportunity to invest in and develop more renewable energy.

Q: You’ve painted a very positive picture. You’ve always been at the forefront of the industry and see greater potential, so I understand your thinking. But what worries do you have about your industry?

Fei-Fei Li: I’m neither a tech utopian nor a dystopian—I’m actually a mediocre centrist. The mediocre centrist wants to look at this issue from a more pragmatic and scientific perspective. Of course, any tool in the hands of someone with ill intent worries me. Since the dawn of civilization, fire has been a crucial invention for our species, but using fire to harm is extremely evil. So any misuse of AI worries me. Improper ways of communicating with the public also worry me, because I do feel there is widespread anxiety.

Fei-Fei Li: My only real concern is our teachers. My personal experience tells me they are the backbone of our society. They are crucial to nurturing our next generation. Are we communicating with them properly? Are we involving them? Are our teachers using AI tools to enhance their professional skills or help our children use AI?

Q: Your life today is worlds apart from your days growing up in a dry-cleaning shop. Are you aware of your power as an industry leader?

Fei-Fei Li: (laughs) I still do my own laundry at home. I’m very aware of the responsibility I carry. I am one of the people bringing this technology to the world. I’m honored to work at one of the world’s top universities, training future leaders and conducting cutting-edge research. I know I’m both an entrepreneur and the CEO of one of the most exciting AI startups. So everything I do has consequences—that’s my responsibility. I take it very seriously, because I always tell people: In the age of AI, the initiative should be in human hands. The initiative doesn’t lie with machines, but with ourselves. The mission of my organization is to create exciting technology and use it responsibly.

Q: So, for the people in your life—your children—what do you not allow them to do with AI, electronic devices, or the internet?

Fei-Fei Li: The old saying: don’t use tools to do stupid things. You have to think about why and how you use a tool. It’s actually simple: don’t be lazy just because you have AI. If you want to understand math, large language models might give you the answer, but that’s not the right way to learn. Ask the right questions. On the other hand, don’t use AI to do bad things. For example, the authenticity of information: fake images, fake voices, fake texts. These are problems AI faces, but also problems brought by our current social media-dominated communication methods.

Q: In this rapidly changing world—one we couldn’t have imagined three years ago—you’re calling for a return to traditional values.

Fei-Fei Li: You can call it old-fashioned, or you can call it eternal. As an educator and a mother, I believe some human values are eternal, and we need to recognize that.