DeepMind’s Project Genie blends creativity and AI experimentation

[etfs]

Project Genie lets users create whimsical, interactive worlds from text or images, though the results can be inconsistent. Worlds based on artistic prompts — like claymation, watercolor, or cartoon styles — performed best, while photorealistic or cinematic prompts often fell short.

Users can remix existing worlds, explore curated galleries, or generate videos of their creations. Currently, sessions are limited to 60 seconds due to compute constraints, but DeepMind sees this as enough to gather feedback. Research director Shlomi Fruchter said the team hopes to improve realism and interaction capabilities in future iterations.

[mendez]

remixing worlds and exploring galleries makes this feel more like a toy than a tool right now, which is fine.