

Announcing Google's Secret New AI Model With The Person Who Built It | Logan Kilpatrick
32 snips Aug 26, 2025
Logan Kilpatrick, Product Lead at Google DeepMind, dives into the revolutionary Gemini AI model, showcasing its remarkable image generation and editing capabilities. He highlights how Gemini enhances creative productivity through advanced features like 'vibe coding.' The discussion further explores AI's potential in robotics, scientific research, and its transformative impact on search technology. With a focus on user accessibility and innovative tools like Google AI Studio, Logan reveals how these advancements are set to reshape our interaction with AI.
AI Snips
Chapters
Transcript
Episode notes
Unified Multimodal Intelligence
- Gemini fuses image generation, editing, and world knowledge to produce grounded, realistic edits.
- The model uses physics and context to avoid blind, implausible image changes.
Decades-Style Photo Edits
- Logan demoed a "Passed Forward" app that regenerates your photo across decades while keeping character consistency.
- He highlighted speed and that the model is Gemini 2.5 Flash Image (Nano Banana) available in AI Studio.
Multimodality Enables Transfer
- Training Gemini natively multimodal enables capability transfer between vision, video, and text.
- This fusion is part of the path the team sees toward more general intelligence.