AI EngineerVideo
Let's go Bananas with GenMedia — Guillaume Vernade, Google DeepMind
5.5Score
Watchable video resourceOpen original video
TL;DR · AI Summary
Google DeepMind developer advocate Guillaume Vernade introduces the vision of GenMedia multimodal models, but the content is primarily personal experience and conceptual descriptions lacking technical details.
Key Takeaways
- DeepMind's ultimate goal is to build a unified world model that processes all mo
- Currently still using specific models like image generation and video generation
- Developer advocate role requires ensuring API design meets developers' actual ne
Outline
Jump quickly between sections.
Guillaume Vernade has worked at Google for 6 years and is currently a developer advocate at DeepMind.
Responsible for ensuring complete documentation, code samples, and prompt guides when products are released.
DeepMind is committed to building world models that can understand and output multiple modalities.
Mindmap
See how the topics connect at a glance.
查看大纲文本（无障碍 / 无 JS 友好）
- GenMedia多模态模型
- 开发者体验
- 文档与示例
- API一致性
- 技术愿景
- 世界模型
- 多模态处理
Highlights
Key sentences worth saving and sharing.
My job is to make sure developers have everything they need: documentation, code samples, demos, and new skills and prompt guides
World model is a model that can ingest as many modalities as possible (sound, video, audio, sensors) and output multiple modalities
For release purposes, it's easier to ship specific models and avoid breaking other functionality when updating the main model
#GenMedia#Multimodal AI#Google DeepMind#Developer Experience