Ashley Edwards is a senior research scientist working on reinforcement learning and foundational world models. She most recently spent time at Google DeepMind where she worked on generating interactive environments from videos. Before that, she was a research scientist at Uber AI working on training agents from observations. She received a PhD in 2019 from Georgia Tech and focused on developing models that can infer latent actions, rewards, and policies from videos alone.
Talk Title: Genie: Generative Interactive Environments
Abstract:
In this talk we’ll discuss Genie, the first generative interactive environment trained in an unsupervised manner from unlabelled Internet videos. The model can be prompted to generate an endless variety of action controllable virtual worlds described through text, synthetic images, photographs, and even sketches. Genie enables users to act in the generated environments on a frame-by-frame basis despite training without any ground-truth action labels or other domain-specific requirements typically found in the world model literature. We’ll further discuss how the model facilitates training agents to imitate behaviors from unseen videos, opening the path for training generalist agents of the future.
RPL Summer School 2026 