The natural resource costs of AI: Why we must be responsible using "free" technology

We all know that AI technology development is accelerating at an incredible pace and disrupting the world as we know it, but reading this recent article in Forbes made me think of the impact that AI also has on our natural resources. 

Disruption should not draw immediate negative connotations; other advances of this magnitude in the past decades have brought huge benefit to society and business alike – from the mobile telephone to the internet. Artificial Intelligence will no doubt do the same in the future, as we have featured in recent articles.

Yet this article’s focus on the significant increase in water consumption at Microsoft and Google as a result of the cooling systems required for AI supercomputers highlights another resource-intensive consequence of technological advancement, beyond the electricity to power them.

ChatGPT captured the collective mainstream imagination earlier this year, inspiring intrigued users to test out the “free” technology with puzzles to solve, essays to write, and jokes to tell.

But at what cost? The Forbes article estimates a series of 5 to 50 prompts in the system consumes around half a litre of water. It also suggests concerns are rising around residential supply levels near to its data centres.

Tracking whether the benefits of AI outweigh the potential costs will be no mean feat, from environmental, societal, and financial perspectives, and responsible use will become critically important as greater accessibility to AI tools and its widespread integration to many facets of everyday life soon becomes the norm.