The oft-used word "sustainability" states an ideal. It would seem to be a simple one, even a low bar. Can the human species sustain itself on this planet? Do our practices sustain our planet's ability to host us?
Three years ago the University of Alberta put the point this way: "Sustainability means meeting our own needs without compromising the ability of future generations to meet their own needs."
At the same time, "artificial intelligence" is on everyone's lips. Nothing about the near future seems more certain than that there will be ever higher levels of AI there, doing ever more and ever-more complicated tasks.
So, let us promote the battle of the finance world buzzwords, by asking: is AI sustainable? What is the ever-increasing dependence of the human race on self-teaching complexes of algorithm going to do to the environment in which we live?
The reason for asking? AI uses up a lot of power. A story that appears recently in MIT News tells us, "The computational power required to train generative AI models that often have billions of parameters, such as OpenAI's GPT-4, can demand a staggering amount of electricity, which leads to increased carbon dioxide emissions and pressures on the electric grid."
Even beyond that, there is water use. Water is used to cool hardware throughout the world of computing. AI in particular is a very water-intensive field, straining both municipal water systems and the local ecosystem. According to a Forbes study a little more than a year ago, AI's projected water usage could hit 6.6 billion cubic meters of the stuff by 2027.
If AI is not sustainable, it'll have to be shut down. That is definitionally clear, presuming only that we do want to sustain ourselves. Re-adapting it to make it sustainable sounds like only a slightly less ambitious goal than shutting it down altogether.

Comments
Post a Comment