Forget Nvidia: Ndea wants to create AI that runs on its own without ‘barriers’

François Chollet, former Google engineer and creator of widely used Python learning methods. Hardhe started with them Goodnew AI research and science lab, along with Mike Knoop, co-founder of Zapier.

In a post on new beginnings websitethe founders describe their goal of combining cognitive awareness, supported by deep learning, and systematic thinking through what they call “integrated programming.”

According to them, this combination will allow AI machines to adapt and innovate beyond what is happening today, leading to Artificial General Intelligence (AGI), which is loosely defined in the AI ​​community as machine intelligence that can surpass humans in economics. and cognitive functions. While writing:

“We need computers that can solve problems and explore a new field, not just use known methods. We need computers that can create new things. The path to AGI is not just adding to existing methods.”

The couple has yet to say whether or not they have received outside funding for the venture or whether they are just starting it with their own money.

It comes months after former OpenAI co-founder and chief scientist Ilya Sutskever, said to have led a small but later wave of chaos in which his co-founder, Sam Altman, also announced a start-up focused on manufacturing”Safe Superintelligence” and $1 billion in private funding.

More than deep learning

While existing deep learning methods are impressive, Chollet and Knoop argue that they are constrained by their reliance on large data sets and their inability to adapt to new applications.

Chollet and Knoop believe that software integration is the key to solving this problem.

Unlike traditional deep learning, which links between data sets, machine learning searches for programs that describe information. This method allows for a large increase in the number of data points.

Combining deep learning with deep thinking for software synthesis can lead to a new concept of AI research.

“Ndea’s mission is to use AGI to achieve unprecedented scientific progress for the benefit of both current and future generations,” he says.

Building a “Factory for Scientific Advancement”

Ndea’s long-term vision extends beyond creating AGI. The lab aims to act as a “rapid scientific development factory,” capable of solving known and unknown problems.

From tackling current frontiers like autonomous vehicles and sustainable energy to accelerating new discoveries, the lab sees itself as a catalyst for the advancement of science.

Chollet added that their research methods have the potential to unlock breakthroughs and redefine the boundaries of human knowledge. As he wrote in a thread on X: “If we succeed, we will not stop at AI. With this technology in hand, we want to solve any scientific problem that it can solve. We see scientific progress as the most exciting way for AI.”

According to Chollet, this progress depends on creating an AI that can learn as well as humans and continue to improve over time without obstacles.

Although he admits that success is not guaranteed, Chollet emphasized the importance of pursuing this main goal, referring to X: “We believe that we have a small but real chance of success – to create an AI that can continue to improve over time without obstacles. .”

New AGI Research

Software synthesis, the cornerstone of Ndea’s research, is still in its infancy. Chollet compared the current situation to what deep learning was like in 2012.

However, he noted that its potential is being widely recognized by AI laboratories, although many see it as only a small part of what is needed for AGI.

Ndea, in contrast, sees software integration as equally important as deep learning and has made it a priority in their approach.

The lab is also recruiting a team of researchers and engineers distributed around the world to create what it describes as the most talented team in the world.

The company operates as a remote organization and is looking for people with strong skills, especially in translating math into code.

Founders with a solid track record

François Chollet and Mike Knoop bring a wealth of experience to Ndea.

At Google, Chollet conducted extensive research on deep learning and AI systems, identifying limitations of existing models and opportunities for improvement. Its offerings include not only Keras but also ARC-AGI benchmark, the most widely used metric to measure AGI progress.

He is also the author of this book Deep Learning with Python and has been recognized among Time’s “100 Most Influential People in AI.”

Knoop founded Zapier, the world’s largest AI company, where he led engineering and product development and early adoption of AI technology.

He is also credited with pioneering the distributed teams around the world. Both Chollet and Knoop are co-founders of the ARC Prize Foundation, a non-profit organization focused on advancing AGI research.

A vision of the future based on ancient traditions

Ndea takes its name from a Greek concept ennoia (natural understanding) and dianoia (ideal reasoning), reflecting the lab’s goal of combining deep learning with software synthesis. By using AGI, Ndea hopes to force scientific progress for decades to years or years.

Despite acknowledging the uncertainties and challenges of pursuing AGI, Chollet and Knoop remain optimistic about their path. He sees AGI as a way to solve the problems people are facing and to reveal new opportunities for discovery.