François Chollet Launches Ndea, a New AI Lab Focused on AGI

François Chollet, one of AI’s most influential voices, has co-founded Ndea, a research lab with a singular focus: building artificial general intelligence (AGI) capable of true invention and innovation. Partnering with Mike Knoop, co-founder of Zapier, Chollet envisions a paradigm shift in how AI systems learn, adapt, and contribute to scientific progress.

Key Points:

Ndea focuses on merging program synthesis with deep learning to create more adaptable AI systemsThe company aims to accelerate scientific progress through AI that can innovate independentlyThe startup is building a distributed team of program synthesis researchers

Chollet’s track record speaks volumes. As the architect behind Keras, a Python deep learning library powering AI development at YouTube, Waymo, and Spotify, he has shaped how developers approach machine learning. His work at Google, spanning nearly a decade, included creating the ARC dataset, a benchmark for measuring AGI progress that has influenced how researchers evaluate artificial intelligence.

The timing of Ndea’s launch is particularly noteworthy. Last week, Chollet and Knoop transformed their $1 million ARC Prize competition into a formal nonprofit organization. The competition’s insights appear to have influenced Ndea’s approach – while companies like OpenAI achieved high scores on the ARC-AGI benchmark, they required extensive computing power and still fell short of human-level performance on tasks that children master quickly.

The ARC Prize will Become a Nonprofit that Benchmarks AGI

The foundation will transition to a 501(c)(3) nonprofit and begin fundraising this month

Ndea’s founding stems from a critical view of current AI limitations. While today’s language models excel at pattern matching and specific tasks, they struggle with open-ended problems and efficient learning. “Current deep learning-based AI, while impressive and economically valuable, is ultimately constrained by its inability to efficiently learn and adapt,” Chollet explained on X (formerly Twitter). “We’re betting on a different path to build AI capable of true invention, adaptation, and innovation.”

The company’s technical strategy centers on program synthesis, a method that searches for discrete programs to explain observed data, potentially requiring fewer examples than traditional deep learning approaches. By combining this with deep learning techniques, Ndea hopes to create systems that can both recognize patterns and engage in formal reasoning. In fact, the name Ndea draws inspiration from the Greek concepts of intuitive understanding (ennoia) and logical reasoning (dianoia), reflecting its dual approach to artificial intelligence.

This launch reflects a growing trend of prominent AI researchers leaving Big Tech to establish independent labs. Chollet joins others like former OpenAI chief scientist Ilya Sutskever in pursuing alternative approaches to AI development. Knoop, meanwhile, is stepping back from day-to-day operations at Zapier to dedicate himself to Ndea, bringing experience in scaling globally distributed teams and AI-driven automation.

Program synthesis, while promising, faces substantial computational hurdles. Ndea believes it can overcome these challenges by using deep learning to guide the program search process. The company likens the current state of program synthesis to deep learning circa 2012 – less mature but showing significant potential, with major AI labs beginning to explore its applications.

Looking beyond AI development, Ndea envisions itself as a catalyst for scientific advancement. The company plans to apply its technology across various scientific domains, positioning itself as what they term a “factory for rapid scientific advancement.”