- Google has published a study describing the amount of energy and water that it uses
- Experts say that the figures lack important context
- Google insists that these figures represent the average user experience
A new Google study claims that its gemini AI model only uses very minimal water and energy for each prompt – with the median use sitting at around 5 drops (0.26 milliliters) – the equivalent electricity used for 9 seconds of television observation (approximately 0.24 watt -heures), which resulted approximately 0.003 grams of CO2 emissions.
Experts quickly disputed the complaints, however, with The penis By saying that Google has omitted key data points in its study, considerably underdeveloped the environmental impacts of the model.
Although models and data centers have become more effective, it seems that there is more in history than Google.
The tip of the iceberg
One of the authors of an article cited in the study, Shaolei Ren, an associate professor of electrical engineering at the University of California, told the publication; “They just hide critical information. It really broadcasts the bad message to the world. ”
AI models and Gemini are supported by data centers – huge warehouses full of servers that consume intense quantities of water and energy, supporting local resources.
Governments around the world have sanctioned the construction of these data centers, despite the destruction they could bring to the local campaign – and consumers are probably those who pay for additional energy used.
One of the largest concerns about Google’s study is that it omits the indirect use of water in estimates, which constitute the majority of use -related use. While the figures are technically correct, the missing context of extreme energy consumption depicts a misleading image.
The study only looks at the water used by the data centers to cool their servers, but it is left aside the electricity that these data centers require, which in turn leads to new gas and nuclear power plants – which also cool their systems with water, or use steam to turn turbines.
Water is not the only distorted Google metric, however, the document describing only a market -based carbon emission measurement, which compensates for the figure using Google’s promises to use renewable energies to support electricity networks. Savannah Goodman, head of advanced energy laboratories Techradar Pro,
“We hope to share environmental measures representative of the behavior of a typical user, and reasonably comparable over time. However, with the rapidly evolving AI model architectures and an assistant user behavior of AI, there are aberrant values either from small subsets of prompts served by models with low use, or with high chip accounts.”
“In order to share metrics that represent the experience of a typical user and are robust in this rapid development area, we have chosen to measure the measures for the median prompt – which is robust at extreme values and provides a more precise reflection of the energy impact of a typical prompt.”
