Overview
Explore the concept of "hardware lottery" in artificial intelligence research through this in-depth video analysis. Delve into how certain algorithmic ideas succeed due to their compatibility with prevalent hardware, while equally viable ideas are left behind. Examine historical examples of hardware lotteries, their impact on research progress, and the current state of AI development. Investigate the disconnect between machine learning researchers and hardware, using GPT-3 as a case study. Compare the scaling of neural networks to human brains and discuss potential ways forward in AI research. Gain valuable insights into the interplay between hardware, software, and algorithmic ideas in shaping the future of artificial intelligence.
Syllabus
- Intro & Overview
- The Hardware Lottery
- Sections Overview
- Why ML researchers are disconnected from hardware
- Historic Examples of Hardware Lotteries
- Are we in a Hardware Lottery right now?
- GPT-3 as an Example
- Comparing Scaling Neural Networks to Human Brains
- The Way Forward
- Conclusion & Comments
Taught by
Yannic Kilcher