On the History and Future of AI
summary
Tells how modern vision AI combines neural networks, GPUs and a curated ImageNet of 50 million images to move from labeling to generative and 3D models; cites diffusion models (Sora) and a Stanford model “Walt” released months earlier. Describes single-image-to-3D and text-to-room layouts, training robots in Nvidia Omniverse simulations and the Behavior project, shows a robotic arm following LLM-style verbal commands and an EEG-controlled robot cooking Sukiyaki, plus pilots for hand‑hygiene, instrument tracking and fall alerts.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 1:50Vision and intelligence dawn
- 3:17AI breakthroughs in vision
- 5:27From images to video generation
- 7:14Why seeing alone is not enough for AI
- 9:03Spatial intelligence in AI
- 11:13Digital world capture
- 12:36Robotic learning
- 14:06AI in healthcare
- 15:55Brain-controlled robots
- 16:40Digital Cambrian explosion
related talks
18mTrinity of Artificial Intelligence
36mArtificial Intelligence: The Future of Software
44mCan AI See You? Your Introduction to Computer Vision and GenAI
17mBeyond Automation: AI and the Evolution of Robotic Intelligence
29mThe Present & Future of AI and Games
19mAI Trends in Intelligent Transport Systems
38mHow Artificial Intelligence has changed your life
03mInterview