Fei-Fei Li: AI policy needs science, not science fiction

In an interview with Emily Chang on Bloomberg's The Circuit, published on August 19 on YouTube, the ImageNet creator and World Labs CEO explained why AI's future lies beyond chatbots — and why AI policy needs science, not science fiction.

The end of the chatbot era

While the industry focuses on language models like ChatGPT and Claude, Li is betting on a different frontier: world models, systems that predict what happens next in the real world — not just the next word in a sentence.

"If we want to create a world where we continue to push scientific discovery, where we want machines like robots to be a partner to us, all this language cannot do it alone. Building world models and spatial intelligence is not about anti-LLM. It's about the next frontier and the next chapter."

What is a world model (and what World Labs builds)

For Li, spatial intelligence breaks down into three functions: rendering, simulation, and planning. World Labs, the startup that raised $1 billion, launched Marble, a platform that generates explorable, editable 3D worlds from a single visual or text prompt.

• Rendering: the model outputs beautiful pixels for humans to consume (like OpenAI's Sora).
• Simulation: captures the actual structure of the world, geometry based on physics — serving machines, not just humans.
• Planning: tells you what to do next — like a robot knowing to pick up a cup and move it somewhere else.

"Marble is no different from a picture — until you realize it's actually a world, because I can navigate in it. I can move forward. I can turn around. I can go anywhere. It's a fully consistent 3D world."

Use cases are already concrete: virtual production in movies, game development, and robot training in partnership with NVIDIA.

Robotics: $6 billion is too small

Asked about the gap between hype and reality of humanoids — which still can't load a dishwasher — Li pushes back: robotics will be one of the most important revolutions in human industrialization, and current investment is insufficient.

"$6 billion is too small. If you look at self-driving car investment, if you look at language model investment, it took way more than $6 billion."

"I'm paranoid every day"

Competing against deep-pocketed giants, Li admits the pressure — but not paralyzing fear. The advantage, she says, is focus.

"As an entrepreneur and CEO, I'm paranoid every day. But am I paralyzed? Absolutely not. Big companies have a lot on the table. We have one thing on the table, and we're focusing on that thing."

Investment in world models already totals $3 billion and growing, but the field is still immature — comparable, Li says, to 2019 for chatbots.

AI policy: science, not science fiction

An adviser to US presidents and the UN, Li has a clear recipe for regulation. She criticizes a debate dominated by extremes — total utopia or total doomsday — and defends evidence.

"First and foremost, root that in science, not science fiction. A lot of the AI policy world conversation with people from Silicon Valley talks about the extinction of humanity, the AGI machine overlord, which distracts the real policy work."

Her second recommendation: ensure a healthy ecosystem, resource the public sector and STEM education from K-12 to K-16.

"Human capital, at the end of the day, is the most important resource of our world. And we want a thriving society of people, not a thriving society of just machines."

Risks, governance, and the "God complex"

Li acknowledges the real risks — misinformation, robots weaponized by bad actors, students using technology as a lazy crutch — but rejects panic and messianism.

"It's dangerous for any individual to think they know better than everybody else. As a CEO, our founding premise is we are here to contribute to society and empower people. It's not our position to make every single decision for people."

For her, the way out is collective governance — without throwing the baby out with the bathwater.

"The governance shouldn't be 'let's stop AI,' because this is a technology that can discover cures for diseases, can empower students and teachers, can help our elderly. So that's the baby here."

The optimism of the woman who gave AI its eyes

At the end, Li reveals her philosophy: the arc of history bends toward benevolence. And she leaves a message about science's role in public debate.

"I don't go in hyping. I don't go in providing extreme rhetoric — either total utopia or total doomsday. It's very important that whoever I talk to knows that I'm a scientist. They need scientific facts. And it's my responsibility to speak my mind, to push back if they're wrong."

Her hope, she says, is in humanity. The same curiosity that took her from Chengdu to Princeton, from ImageNet to World Labs, now points to the next leap: machines that don't just see, but understand and navigate the physical world.