Quick Notes:

• increased energy demand

• more fossil fuel / natural gas burning

• cooling systems can cause water depletion

Because AI uses such a large amount of energy (particularly models that generate media), data centers are increasingly devoting more energy for AI. This is not necessarily bad. However, with more demand, this means we are overall increasing the amount of energy we need to create -- often through burning fossil fuels or natural gas. Dr. John Fleming from the Center of Biological Diversity warns that this could potentially increase our reliance on fossil fuels right when we need to move away from them most. In order to maintain mindful progress, we must invest in clean energy infrastructure to safely support the future. 

Unfortunately, right now, data center cooling systems also pull water from the surrounding ecosystem which can come at the cost of aquatic creatures. According to the Washington Post, a 100-word email generated by an AI chatbot requires a little more than a bottle of water; but this builds over time. If one out of ten working Americans has AI write emails for them once a week throughout one year, the water usage eventually equates to Rhode Island's entire household water comsumption for 1.5 days. 

Quick Notes:

• it is up to us

In the status quo, 14% of Google’s freshwater withdrawals for data centers come from sources at high risk of water depletion or scarcity; this water partly goes towards AI. Yet it is important to note that, conversely, 86% of their water use is from low or medium risk areas. 

Like every facet of humanity and human development throughout history, AI is nuanced. The physical impacts of data centers on the immediately surrounding environmental and the global biosphere are latent and current threats. However, this does not mean that AI is a net harm.

For example, it holds remarkable potential for research in many fields and environmental protection is one of them. AI can help us mitigate carbon emissions by improving efficiency in high-impact sectors. It can help us redesign energy infrastructure to become faster, cleaner, and greener. It can also help us prepare for climate adaptation and ecosystem protection as biodiversity and air quality monitoring as well as natural disaster response become ever more critical. 

Overall, it is up to us to choose whether AI will be a force for good or bad in our world. Education about its environmental effects is the first step to making the right choice. 

Quick Notes: 

• see the "Tools for AI Users" tab

• be mindful about your own usage

• stay up-to-date

There are many digital tools that exist, not only track AI's environmental impacts, but to mitigate them as well. AI Energy Score provides a detailed ranking of over 100 open source AI models' energy efficiency while Code Carbon tracks your own individual performance. You can also use Ecosia's AI feature for green, clean use or purify Claude with Carbonlog, a code plug-in that takes only 30 seconds to install!