- Future Blueprint
- Posts
- Google’s Genie generates worlds 🧞
Google’s Genie generates worlds 🧞
Google DeepMind is scaling up what AI can see, understand, and build
Hi AI Futurists,
Google DeepMind just demonstrated Genie 2, a generative AI that can turn a single image into an interactive 3D world for training AI agents and robots. Combined with Astra’s real-world sensing, Google is edging toward AI systems that can learn from both simulation and reality. Let’s take a look.
Here’s our agenda.
Our sponsor: Notion
Google’s Genie generates worlds
Top 3 selected AI tools
AI in the wild
Top news on the AI horizon
Best,
Lex
Manage your settings: Share | Unsubscribe | Upgrade

Get Over $6K of Notion Free with Unlimited AI
Running a startup is complex. That's why thousands of startups trust Notion as their connected workspace for managing projects, tracking fundraising, and team collaboration
Apply now to get up to 6 months of Notion with unlimited AI free ($6,000+ value) to build and scale your company with one tool.

How AI is Impacting the World
Google’s Genie generates worlds 🧞
Google DeepMind is scaling up what AI can see, understand, and build. In a new segment of 60 Minutes Overtime, Google showcased Genie 2, DeepMind's latest generative AI model capable of transforming static images into fully explorable 3D worlds. These simulated environments don’t just look cinematic—they're interactive spaces that AI agents (and eventually robots) can move through, reason in, and learn from. As DeepMind CEO Demis Hassabis explained, “The Genie world model is creating the world around it on the fly.”
Alongside Genie 2, they also tested Project Astra, a multimodal AI assistant that interprets the real world through audio and video input. Astra identified artwork and spun up emotional narratives about the subjects, showcasing a kind of proto-creativity. DeepMind is building AI agents that imagine, generate, and act. And they're training these agents in photorealistic simulated worlds rather than in expensive, slow real-world labs. “You’d learn first in simulated worlds… and then fine-tune in the real world,” said Hassabis.
With tools like Genie 2, DeepMind is creating virtual testbeds that could revolutionize how machines learn. Hassabis hinted that existing geographic data—like Street View—could become interactive landscapes for agents to navigate. If AI can explore any image or location as a 3D environment, it changes what it means to “train” a model. The idea of world-building could shape how future agents learn to think, move, and act long before ever touching the physical world.
Watch the full conversation here.
Key Points
Genie 2 builds interactive environments from static images
These environments are used to train AI agents and potentially robots
Training in simulated worlds reduces cost and time
Astra can now interpret surroundings via audio-visual input
DeepMind may use Google’s geographic data (Maps, Street View) for world-building
Veo 2, DeepMind's video AI, now generates photorealistic, film-like clips from text prompts

What We Think About It
This is a major leap in embodied AI and simulation-based learning. Genie 2 unlocks real scalability for training agents without the physical-world friction. Using static images or geographic data as training environments will accelerate multimodal agent development faster than expected.

What You Can Do Right Now
Start paying closer attention to how AI is showing up in everyday tools—from voice assistants to smart glasses—because versions of Astra will likely become as common as smartphones. Explore AI-powered creativity apps like image generators or video tools (e.g. Runway, ChatGPT’s DALL·E, or Adobe Firefly) to get hands-on with how generative models are reshaping art, storytelling, and communication.

Should 3D worlds be used for more than just gaming? |

Apply AI Superpowers with Tools
AI voice platform for ultra-realistic text to speech and voice cloning
Transform images into 8 second videos
AI assistant that manages your work calendar

AI in the wild
Edit objects in your videos using AI
🚨 Pika just dropped its craziest feature yet — Pika Twists 🔥
Move, stretch, or transform any object in your video, without touching the rest.
Here are 7 mind-blowing examples you need to see 👇
— Devin AI News and Productivity Tools (@toolandtea)
4:45 AM • Apr 16, 2025

AI Quick Hits
Famed AI researcher launches controversial startup to replace all human workers everywhere
Humanoid workers and surveillance buggies: ‘embodied AI’ is reshaping daily life in China
Sam’s Club phasing out checkouts, betting big on AI shopping
OpenAI’s o3 AI model scores lower on a benchmark than the company initially implied
Artificial Intelligence Solves One of Archaeology’s Greatest Puzzles
ChatGPT will now use its ‘memory’ to personalize web searches
Instagram tries using AI to determine if teens are pretending to be adults

What type of coverage would you like to see most? |

That’s all for today, folks!
If you’re enjoying the newsletter, share with a friend by sending them this link: 👉 https://www.futureblueprint.xyz/subscribe
Looking for past newsletters? You can find them all here.
Working on a cool A.I. project that you would like us to write about? Reply to this email with details, we’d love to hear from you!
What do you think about today's edition?We read your feedback every time you answer. |
Reply