By Kaelyn Delaurier, Senior Reporter
As artificial intelligence develops, there are many arguments both for and against the use of this technology. Some note the dangers it poses, from anti-intellectualism to replacement of jobs and stealing creative works. Others acknowledge the usefulness of AI, especially in the medical and research fields. This indecision about whether AI is good or bad surrounds all discussions about the technology, even and especially its impact on the climate crisis.
Now, it’s important to note before anything else that it is difficult to pinpoint the exact effect of AI on climate change, as corporate secrecy and the rapid evolution of AI programs make it difficult to collect accurate data. However, most of us have probably heard about the damaging energy and water requirements of AI. Before the major incline in AI usage, computing used roughly 1-2% of the U.S.’s energy consumption, but that has increased to 4-5% with the new demands of AI. This number could get up to 22% by 2028, which is a massive leap in only three years. Furthermore, energy hit a point in the mid 2000s where its usage remained relatively level, giving companies and researchers the opportunity to switch to more climate-safe energy options. Now, with AI’s rapid demand of energy, this cushion to figure out climate-safe energy solutions is diminishing.
AI’s impact on climate change does not begin and end with the need for energy and carbon emissions. Fifteen ChatGPT queries alone require about half a litre of clean water to cool their systems. Additionally, there are many carbon emissions to factor in beyond the running of the AI processors, such as the emissions required to make the hardware, the damage of mining rare earth minerals for the microchips, and shipping emissions. As if emissions and water usage weren’t enough, due to the rapid change in AI technology, companies constantly need to upgrade their hardware, leaving piles of electronic waste in their wake.
Fortunately, companies are not unaware of the damage AI poses to the climate. Many companies limit the power each processor can draw, reducing the energy workload by up to 20%. They also have turned away from air conditioning to cool their processors, which cuts down on their energy usage and carbon emissions. Unfortunately, this is where AI needs freshwater, since they use water to cool the systems instead of air conditioning. Some companies are looking into the possibility of using a liquid like mineral oil instead of freshwater to cool their systems, but this idea is still in the developmental stages. Some companies also use clean sources of firm energy, such as geothermal and nuclear, since typical renewable energy such as solar and wind power do not provide enough energy for AI systems. Yet, despite the attempt of some companies to mitigate the impacts of AI on the environment, some companies do not care at all and instead are using the easiest form of energy – non-renewable resources – to run their processors for the sake of a quicker profit.
AI is not all bad for the climate crisis. We can use AI to track the climate crisis, from predicting weather patterns to monitoring coral reef health to optimizing supply chains and much more. These technological systems help to efficiently record data and apply this to warning systems or provide better forms of research. On that same note, though, AI can also provide false information, both when recording data and when people look for general information on the climate crisis. These forms of technology are also at risk of cyber attacks, which can cause even more misinformation or leaking of important information.
Thus, instead of simply asking whether AI is or is not good for the environment, we should be asking how we can minimize AI’s damaging impacts on our world. While it seems like only big companies can truly make a difference in this matter, there are small, everyday tasks we can do to minimize AI’s destruction of the climate. Before using AI, ask yourself whether your use of AI justifies the energy it consumes. Also, try to use AI technology during the ‘off hours’ – a time when people are using less energy (such as in the evenings) – like you would other electrical appliances. You can also tighten your AI prompts to ensure you only have to ask AI one or two questions instead of fifteen, and you can opt-out of AI in general searches where possible, such as on Google.
Yes, AI poses many new challenges and threats to our world. It also provides many promises in technological advancement. Until more concrete rules and regulations are created surrounding this new technology, the best we can do is learn to coexist with AI and try to use it as responsibly as possible.
Sources:
Climate Portal “Is AI’s energy use a big problem for climate change?” Is AI's energy use a big problem for climate change? | MIT Climate Portal
Yale Climate Connections “What you need to know about AI and climate change” by Daisy Simmons What you need to know about AI and climate change » Yale Climate Connections