How This AI Entrepreneur is Building Agents That Anticipate the Unexpected

To achieve this, Hafner relies on model-based reinforcement learning. He develops “world models”—AI systems designed to emulate physical reality—and trains agents within these virtual environments. The agent treats the model as a realistic simulation, learning how to act within it. It then leverages these experiences to predict future outcomes—essentially dreaming or imagining what might happen. This capability allows agents, or the robots they power, to navigate unfamiliar situations in the real world (IRL).

“I get to interact with a lot of really smart people in research at Google, and he easily sits in the top half of 1%.”

Timothy Lillicrap, Google DeepMind

Unlike other approaches, Hafner’s technique enables agents and the robots they control to execute highly complex tasks without the traditional real-world trial-and-error training typically required in robotics.

Hafner grew up in a rural town in northeastern Germany, where his parents were both classical musicians. He learned programming from a neighbor, and during high school, he began taking online courses in AI, which quickly turned into a passion. “I was always fascinated with how thinking works,” he says. AI offered him a way to emulate cognitive processes on a computer.

In 2015, as a second-year undergraduate studying engineering at the Hasso Plattner Institute in Potsdam, he secured a role as a student researcher at Google Brain. From there, he went on to hold a dozen internships and other positions across the company, including stints with Google Brain and Google DeepMind (which have since merged) in the UK, Canada, and the US. He worked alongside industry legends, including Geoffrey Hinton—often called one of the “godfathers” of AI—and Ashish Vaswani, co-author of the groundbreaking paper “Attention Is All You Need,” which introduced the transformer technology underlying today’s large language models.

Source link