BETA RELEASE

Summary

A Q&A with UW professor Sajjad Moazeni regarding the energy consumption of training and running large language models like ChatGPT.

Key quotes

Overall, this can lead to up to 10 gigawatt-hour (GWh) power consumption to train a single large language model like ChatGPT-3.
This many queries can cost around 1 GWh each day, which is the equivalent of the daily energy consumption for about 33,000 U.S. households.

The article features Sajjad Moazeni, an assistant professor of electrical and computer engineering. He explains the infrastructure requirements and energy costs associated with AI data centers.