Whenever there is a prompt given to ChatGPT, it runs through servers to find out the best responses. Servers generate heat and water cools them down through cooling towers. The amount of water and electricity a data center requires usually depends on where the data center is located. Some data centers also use large units resembling air conditioners to cool down data centers. But even if a data center uses water or electricity to cool down, it still affects the environment.
The water required for AI chatbots using GPT-4 to generate a 100-word email varies by data center location. For example, Washington consumes 1,468 milliliters, while Texas needs just 235 milliliters. This variation emphasizes the environmental impact of technology. Here’s a breakdown of water usage: Washington—1,468 ml, Arizona—925 ml, Illinois—464 ml, Iowa—462 ml, Wyoming—369 ml, Virginia—353 ml, and Texas—235 ml.
AI chatbot water usage varies by data center location, with Washington requiring 1,468 milliliters per email. |
Shaolei Ren, an associate professor at UC Riverside, says that if a data center is located in a hot region, it requires a lot of energy to cool down. If an air conditioner is used for cooling, it takes too much electricity. If water is used for cooling down, it can decrease the water level in that region especially if it is a drought prone area.
Northern Virginia has the world’s most data centers and people often protest that they are building more data centers there and there are still no long term jobs present. Records from Iowa show that Microsoft has already used the region's 6% of total water. Google disclosed that it has used over a quarter of water from The Dalles for their data centers.
According to AI experts, chatbots need months of training and huge amounts of energy before they can answer prompts. During the training process, each chatbot gives out heat in data centers but many bit tech companies like Microsoft, Google and Meta have pledged to use new cooling methods that wouldn't be harmful to the environment. But these pledges are often not fulfilled. Google recently shared a report about its carbon emissions stating that their carbon emission footprint has risen 48% because of their AI data centers. AI is demanding more energy from data centers which can have adverse effects on the environment and the future of this world.
Read next: From Code to Crisis: The Startling Energy Consumption of Top Tech Giants Revealed!
by Arooj Ahmed via Digital Information World
No comments:
Post a Comment