A new artificial intelligence startup aims to move beyond chatbots and build machines that understand the real world.
A leading artificial intelligence researcher has launched a bold effort to challenge the technology industry’s growing focus on chatbots.
According to Quartz, Yann LeCun, former head of AI research at Meta Platforms, has raised approximately $1.03 billion for a new startup called Advanced Machine Intelligence (AMI).
The startup aims to explore a different direction for artificial intelligence one that moves beyond today’s popular chatbot systems powered by large language models.
A Different Path to AI
Over the past few years, major technology companies have invested heavily in large language models (LLMs), the technology behind tools like ChatGPT. These systems can generate text and answer questions but primarily work by predicting the next word in a sequence.
LeCun argues that such systems do not truly understand the world. Instead, AMI plans to develop “world models,” AI systems designed to understand physical environments, cause-and-effect relationships, and how events unfold over time.
His research framework for this idea is called JEPA (Joint Embedding Predictive Architecture), which aims to help machines learn deeper patterns and relationships rather than simply generating convincing text.
Why It Matters
World-model AI could be particularly useful in industries where errors carry serious risks, including manufacturing, robotics, and healthcare. In such fields, incorrect answers—often called “AI hallucinations”—can have costly consequences.
By focusing on real-world reasoning and planning, AMI hopes to create systems that make more reliable decisions.
Leaving Meta to Pursue the Vision
LeCun spent more than a decade building the AI research division at Meta before leaving the company in late 2025. As tech firms raced to develop bigger chatbots, he chose to pursue a different path.
He said developing the technology independently would allow his team to build it “faster, cheaper, and better.”
AMI’s chief executive, Alexandre LeBrun, said the company is not rushing to release products. Instead, it plans to focus on deep research—even if meaningful results take years.
Growing Interest in World-Model AI
AMI is not alone in exploring alternatives to chatbots. Recently, AI researcher Fei‑Fei Li raised roughly $1 billion for World Labs, which focuses on “spatial intelligence” that enables machines to understand three-dimensional environments.
Meanwhile, chip giant Nvidia has also described world models as crucial for training robots and autonomous vehicles.
Major Investors Back the Bet
AMI’s funding round attracted support from major investors including Samsung, Toyota Ventures, and Temasek Holdings.
Individual investors include billionaire entrepreneur Mark Cuban and former Google CEO Eric Schmidt.
The company is headquartered in Paris, with planned offices in New York City, Montreal, and Singapore.
A High-Risk Vision
Despite its massive funding, AMI has no product on the market yet and is not promising quick results.
But if LeCun’s vision proves correct, the next major wave of artificial intelligence may come from machines that understand the world itself—not just the words typed into a prompt.
About The Author
