2021/11/16
论文一
论文二
论文标题:The Computational Limits of Deep Learning
作者:Neil C. Thompson, Kristjan Greenewald, Keeheon Lee, Gabriel F. Manso
期刊:arXiv
发表时间:2020/07/10
数字识别码:arXiv:2007.05558
摘要:Deep learning's recent history has been one of achievement: from triumphing over humans in the game of Go to world-leading performance in image recognition, voice recognition, translation, and other tasks. But this progress has come with a voracious appetite for computing power. This article reports on the computational demands of Deep Learning applications in five prominent application areas and shows that progress in all five is strongly reliant on increases in computing power. Extrapolating forward this reliance reveals that progress along current lines is rapidly becoming economically, technically, and environmentally unsustainable. Thus, continued progress in these applications will require dramatically more computationally-efficient methods, which will either have to come from changes to deep learning or from moving to other machine learning methods.
(Google English translation: )
AI is getting stronger and stronger, but we are about to be unable to support it
The development of artificial intelligence is somewhat difficult
2021-12-3 Hantao Gu Tokyo
Scholars originally predicted and hoped that from about 2010, 35 years later, the development of science, about 2045, artificial intelligence will surpass the human brain and solve various medical problems. About 35 years later, artificial intelligence will be further developed, and it will take twice as long, that is, 70 years later (or 70-100 years later), about 2080 to 2110, to plan and guide implementation, etc., to achieve cancer-free , Do not suffer from dementia, can engage in light work and basic life, self-care, the longest healthy life expectancy is 157-214 years old and so on. Our human brains are already a bit tired and tired. The original hope is artificial intelligence, but it can be a little easier. However, the unfavorable news came, it may be more difficult to bear and realize because of the excessive energy consumption and high cost of artificial intelligence. It's a bit of a pity and a pity.