Assertion Supported AI assessment confidence: 90% certainty 4/5 debate potential 2/5

Li: Current AI cannot independently derive Newton's 17th-century laws of motion

Dr. Fei-Fei Li · The Godmother of AI on jobs, robots & why world models are next | Dr. Fei-Fei Li · Nov 16, 2025 · at 29:26

Dr. Fei-Fei Li argues that current AI lacks true scientific abstraction and creative discovery capabilities.

0:00 / 0:22exact quote · 22.1s

▶ Watch the full episode on YouTube → 720p mp4 · rendered on demand · StarZero watermark

“Let's give AI all the data, including modern instruments data of celestial bodies, which Newton did not have, and give it to that, and just ask AI to create the sixth, 17th century set of equations on the laws of bodily movements. Today's AI cannot do that.”

quote is from the automated transcript, cleaned for reading: filler sounds and stutters are removed, nothing is rephrased. names can be misheard (the analysis reads context, assessments check outside sources). how →

The claim, plainly

Current AI models given modern celestial body data cannot independently derive Newton's 17th-century laws of motion and governing equations.

a quote alone can be ambiguous, so this restates it with the references and context of the conversation around it resolved. AI-written; the tape below the player is the source.

Assessment

Dr. Fei-Fei Li's claim that AI systems in November 2025 could not independently derive Newton's 17th-century laws of motion and universal gravitation from raw celestial data is fully supported by the scientific literature. While modern machine learning and symbolic regression frameworks (such as Lemos et al., 2023, and AI-Descartes, 2023) have recovered gravitational equations from orbital data, they rely on pre-programmed physical priors, such as hard-coding Newton's second and third laws into the graph neural network architectures, providing predefined axiom libraries, or using human-guided multi-stage pipelines. Furthermore, studies evaluating foundation models on orbital trajectories (such as Vafa et al., ICML 2024) confirm that models learn empirical curve-fitting rather than extracting underlying causal Newtonian mechanics without explicit architectural scaffolding.

Lemos et al. (2023) required Newton's second and third laws as hard-coded inductive biases within GNN architectures to infer the gravitational force law from Solar System data.
Axiomatic and symbolic regression discovery systems (AI-Descartes, 2023; AI-Hilbert, 2024) depend on predefined domain axioms, formalisms, and operator libraries rather than raw autonomous derivation.
Vafa et al. (ICML 2024) showed that transformer models trained on celestial body trajectory data perform pattern-matching/Keplerian interpolation but fail to extract underlying governing force laws across counterfactual configurations.

Checked against

[1] arxiv.org [2] arxiv.org [3] qudata.com [4] qeios.com [5] arxiv.org [6] researchsquare.com [7] mit.edu [8] thealgorithmicbridge.com

"AI" discover "Newton's laws" OR "law of gravitation" celestial trajectory data AI rediscover Kepler Newton laws celestial data symbolic regression rediscovering Newton's laws from observational data AI machine learning AI Descartes "Newton" celestial Kepler symbolic regression

assessed as of 2026-09-03

This assessment is an AI reading of a public statement, made against the cited references with the confidence shown. It can be wrong. The quote, the tape and the references are all on this page; if something is off, use report an error below. full disclaimer →

Mentioned

AI Newton's laws of motion

⚑ Spot an error? Report it

More from Dr. Fei-Fei Li

Opinion

Li: AGI is more of a marketing term than a scientific one

“I feel AGI is more a marketing term than a scientific term. As a scientist and technologist, AI is my North Star, is my field's North Star, and I'm happy people call it whatever name they want to call it.”

Dr. Fei-Fei Li Nov 16, 2025 ▶ 25:48 The Godmother of AI on jobs, robots & why world models are next | Dr. Fei-Fei Li

Opinion

Dr. Fei-Fei Li: Scaling laws alone are not enough; AI needs innovation

“I think scaling laws of more data, more GPUs and bigger current model architecture is there's still a lot to be done there, but I absolutely think we need to innovate more.”

Dr. Fei-Fei Li Nov 16, 2025 ▶ 26:48 The Godmother of AI on jobs, robots & why world models are next | Dr. Fei-Fei Li

Insight

Li: Self-driving cars on 2D surfaces are simpler than 3D robotics

“And self-driving cars are much simpler robots. They're just metal boxes running on two D surfaces. And the goal is not to touch anything. Robot is three D things running in three D world. And the goal is to touch things.”

Dr. Fei-Fei Li Nov 16, 2025 ▶ 46:22 The Godmother of AI on jobs, robots & why world models are next | Dr. Fei-Fei Li

Opinion

Li: Spatial intelligence and world models are as important as LLMs

“We believe that spatial intelligence and world modeling is As important, if not more, to language models and complementary to language models.”

Dr. Fei-Fei Li Nov 16, 2025 ▶ 48:43 The Godmother of AI on jobs, robots & why world models are next | Dr. Fei-Fei Li

Assertion Not checkable as stated

Li: World Labs created the first generative model outputting genuine 3D worlds

“We've spent a year and plus Building the world's first generative model that can output genuinely three D worlds.”

Dr. Fei-Fei Li Nov 16, 2025 ▶ 49:16 The Godmother of AI on jobs, robots & why world models are next | Dr. Fei-Fei Li

Assertion Contradicted

Li: World Labs runs real-time 3D video generation on a single H100

“Just a couple of weeks ago, we rolled out the world's first real-time demoable real-time video generation on a single H 100 GPU.”

Dr. Fei-Fei Li Nov 16, 2025 ▶ 59:41 The Godmother of AI on jobs, robots & why world models are next | Dr. Fei-Fei Li

Li: Current AI cannot independently derive Newton's 17th-century laws of motion

More from Dr. Fei-Fei Li

Turn any episode into a week of clips.