New venture aims to build AI that understands the real world

New startup Advanced Machine Intelligence aims to develop AI that learns from real-world data.

Biased AI writing assistants can subtly shift users’ opinions, even when warnings are given, affecting views on societal issues.

AI pioneer Yann LeCun has secured more than $1 billion in funding for a new startup that aims to rethink how AI systems learn about the world.

The venture, called Advanced Machine Intelligence (AMI), will focus on developing AI that learns from real-world signals, such as camera and sensor data, rather than relying primarily on text. According to this French company, such systems could make better decisions by understanding how events unfold in the physical world.

AMI plans to build what researchers call ‘world models’, AI systems designed to predict the consequences of actions before they happen. Developers believe that grounding AI in real-world data could make the technology more reliable and easier to control, especially in critical safety applications.

Operations will span several global research hubs, including Paris, New York City, Montreal and Singapore. The company has already begun assembling its leadership team, appointing entrepreneur Alex LeBrun as chief executive and AI researcher Saining Xie as chief science officer.

Support for the project quickly appeared online. Emmanuel Macron welcomed the launch, saying it represented a new chapter in AI and highlighting the role of researchers and innovators in shaping the technology’s future.

LeCun is widely regarded as one of the key figures behind modern AI. In 2018, he shared the prestigious Turing Award with fellow researchers Geoffrey Hinton and Yoshua Bengio for their contributions to deep learning.

Research at AMI will focus on building AI systems that can reason, plan actions and maintain long-term memory. Possible applications range from robotics and industrial automation to healthcare and wearable technologies, areas where dependable AI could have a major impact.

LeCun and his team argue that genuine intelligence cannot emerge from language alone. Understanding the world, they say, requires machines that learn directly from it.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!