Compute and Breakthroughs
Lex Fridman asked Ilya Sutskever if there are still beautiful and mysterious properties and neural networks that are yet to be discovered. Ilya Sutskever responded that they are still massively underestimating deep learning, and that progress will continue to be made for quite a while. He discussed how individual researchers may find it harder to make breakthroughs due to the large number of researchers, and that compute and large efforts will likely be necessary for breakthroughs. He also discussed the possibility of breakthroughs that may not need huge compute.In this clip
From this podcast
Lex Fridman Podcast
Ilya Sutskever: Deep Learning | Lex Fridman Podcast #94
Related Questions