Lennart Heim on the AI Triad: Compute, Data, and Algorithms
Future of Life Institute Podcast
00:00
The Importance of Algorithmic Efficiency in Machine Learning
If we had optimal algorithms, how much would this matter? For example, I saw a DeepMind recently made an advance in algorithmic matrix multiplication. Could this make say compute less relevant because algorithms would be so efficient that they would need less compute to run? There's definitely history. But again, compute trends have been like a bit faster, right? Because eventually it's not about achieving capability X, it's about like you wanna keep on pushing a frontier, right?
Transcript
Play full episode
