Many of us have been quick to adopt the latest AI tools, particularly in the workplace where we’re interacting with generative AI on an almost daily basis – but I wonder how many of us have given thought to the dark secret behind this time-saving tech.
Much like cloud computing and cloud storage, running the computers that train artificial intelligence requires almost immeasurable quantities of resources, and it seems that one of its most egregious consumptions is water.
Is AI bad for the environment?
While the computers that form the foundation of the AI tools we know and love today are incredibly power-hungry, much work has already gone into making their components as efficient as possible while eking out the maximum performance, and that work continues.
Something that is harder to alleviate is the heat generated by the hardware, which requires constant cooling. In the case of data centers and HPC across the globe, water cooling remains one of the most common and effective methods of keeping temperatures low, but this comes at a cost, often putting local populations (of both people and wildlife) and risk and in some cases, causing the rerouting of water courses.
Microsoft’s very own reporting shows a 34% year-on-year increase in water consumption for 2022, and with its rapid investment in AI in recent months, this is a figure that could grow even further by the time Redmond releases its next report. Google also noted a 22% increase in its water consumption for 2022, similar to the year prior to that, too, but up from 10% for 2020.
According to The Associated Press, research is underway into the environmental impacts associated with using ChatGPT, and has already found that the chatbot uses approximately 500ml of water to process between five to 50 prompts.
TechRadar Pro asked OpenAI, the company behind the headline-grabbing ChatGPT, to share its thoughts on the environmental situation caused by using such technologies, but the company did not immediately respond.
In the meantime, refraining from using power-hungry resources when they are not totally necessary is a responsible way for businesses and their workers to play their own role in the cutback.
More from TechRadar Pro