Training a single AI model can emit
as much carbon as five cars
 |
| IT GETS LESS THAN A THOUGHT PER GALLON |
Deep learning has a terrible carbon footprint, writes
Karen Hao.
The impact: AI researchers have long suspected that deep learning has a sizeable environmental impact, but new research quantifies just how bad it is. Training a single common, large AI model can emit over 626,000 pounds of carbon dioxide equivalent.
How? Advances in AI require ever larger models on sprawling data sets. This approach is computationally expensive—and highly energy intensive.
The significance: It’s colossal, especially when considering the current trends in AI research.
Read the full story here.
 |
| PUE is the ratio of data center total energy used to the energy delivered to computing equipment |
