Google DeepMind Lead Researchers on Genie 3 & the Future of World-Building
The a16z Show
In this episode of the a16z Podcast, Erik Torenberg hosts Shlomi Fruchter and Jack Parker-Holder from Google DeepMind, the team behind Genie 3, along with Anjney Midha, Marco Mascorro, and Justine Moore from a16z, to discuss the groundbreaking world model. The conversation explores Genie 3's ability to generate interactive, persistent worlds from text prompts in real-time, its special memory feature that maintains world consistency, and its potential applications in gaming, robotics, education, and agent training. The speakers also delve into the development process, the emergent behaviors observed in the model, and the future directions of world models, including the possibility of creating multi-player experiences and enhancing realism for applications like robotics training and overcoming personal fears through simulation.
Part 1: Introduction to Genie 3
Part 2: Capabilities and Differentiation
Part 3: Future Applications and Access
Part 4: Trajectory and Philosophical Implications
Sign in to continue reading, translating and more.
Open full episode in Podwise