LeCun thinks AI is being designed incorrectly. And he’s ready to act on what’s right.
The AI industry has spent the last few years chasing one idea: bigger models.
More data. More GPUs. Larger language models.
Yann LeCun thinks that path is wrong.
The former Meta chief AI scientist has launched a new startup called Advanced Machine Intelligence (AMI) and raised $1.03 billion to pursue a different approach to artificial intelligence.
The premise is simple. Current AI systems are effective at predicting text, images, and code. But that does not mean they understand the world.
LeCun argues that today’s large language models cannot produce truly intelligent systems on their own. They generate convincing responses, but they struggle with reasoning, planning, and understanding physical environments.
AMI is trying to fix that.
The company is building AI around what researchers know as “world models.” These systems try to understand how the physical world works rather than predicting the next word in a sentence.
The goal is actually practicality.
Manufacturing, aerospace, and pharma function on complex systems. AI that can reason in real-world environments would greatly manage factories, logistics, robotics, and develop scientific research.
Consumer applications may follow later. LeCun has already suggested that this kind of AI could eventually power hardware like domestic robots or smart glasses.
The timing of the startup is also interesting.
While most AI companies are doubling down on scaling language models, LeCun is betting the industry is heading toward a technical wall. His view? Real intelligence will require systems that understand space, physics, and cause-and-effect relationships. It’s not limited to generation- but a certain understanding of how the world truly operates. And how the world came to be.
In simple words, the next leap in AI will not come from making models bigger.
It might come from making them think differently.
Whether that bet pays off is still uncertain. And with more than a billion dollars behind it, AMI just ensured the AI race now has two competing visions of the future.
