ALL INTELLIGENCE
OpenAI 官方博客T1·Models

OpenAI: Algorithmic Progress Halves Compute Needs Every 16 Months

Original title: AI and efficiency
68AI Score

OpenAI released an analysis showing that since 2012, the compute required to train neural networks on ImageNet has halved every 16 months. This algorithmic efficiency gain (44x reduction) significantly outpaces Moore's Law hardware improvements (11x), suggesting software progress drives more value than hardware scaling in high-investment AI tasks.

Full Report

We’re releasing an analysis showing that since 2012 the amount of compute needed to train a neural net to the same performance on ImageNet classification has been decreasing by a factor of 2 every 16 months. Compared to 2012, it now takes 44 times less compute to train a neural network to the level of AlexNet (by contrast, Moore’s Law would yield an 11x cost improvement over this period). Our results suggest that for AI tasks with high levels of recent investment, algorithmic progress has yielded more gains than classical hardware efficiency.