Considering Artificial Intelligence and Its Potential for Climate Impact
There is an ongoing conversation surrounding the possibilities of Artificial Intelligence (AI) to help us improve energy efficiency and combat climate change. In particular, AI can help spur better climate predictions (due increased pattern recognition abilities), simulate the effects of extreme weather, measure carbon outputs, and “decouple economic growth from rising carbon emissions and environmental degradation.”
This last point emphasizes the ability of Artificial Intelligence to change how, where, and why humans do business. To put it bluntly, the creation of Artificial Intelligence has the potential to be the greatest change to human life maybe ever, and definitely since the Industrial Revolution. However, the ethical question researchers today are tasked with is this—AI has the long-term potential to reduce the effect of humans on the environment, but its creation is strongly entrenched in the fossil fuel industry now. Is increased pollution now worth potential solutions in the long run?
Of course, this question is one of balance. An attuned reader may reply—well, it depends on how big of a potential decrease in climate emissions AI would provide and how much it is contributing now. There is evidence that training an AI system using a deep neural net (a technique used in most high-profile AI systems) takes 626,000 pounds of carbon dioxide.
This amount of carbon dioxide is equivalent to the manufacturing of five cars. However, these are upfront costs, usually necessitated in the training stages of AI formation, not necessarily their day-to-day use. Still, Swedish researchers theorize that by 2030, 6% of the world’s energy will be used to run data centers. These data centers are not AI in and of themselves, but the data they store will help power AI, among other technology.
Additionally, there are problems that supersede just the simple manufacturing of these products. Yes, these systems have the possibility to be used for climate change mitigation, but they can also be used for climate change exacerbation.
Particularly in the U.S., big oil and big tech are increasingly co-involved, with companies like Amazon making AI products that aid in the discovery of fossil fuel. In 2020, Google announced that it would stop this practice, but that was after making $65 million from the industry in 2019.
Finally, AI researchers must consult with philosophers and environmental historians on the implications of potentially creating another sentient, human-like intelligence. If researchers are successful in creating an artificial system that is truly as intelligent (or more intelligent) than humans, they are increasing the complexity and potential disruption of ecological systems on earth.
There is an acknowledgement across the scientific community that humans are a cause of current climate change, beginning in the Industrial Revolution with the increased use of fossil fuels. However, the Industrial Revolution has deeper anthropological roots, originating with the first sign of human agriculture.
It is this evolution towards greater intelligence that allowed early humans to begin forming societies, cities, hierarchies, printing presses, factories, telephones, cars, iPhones, and even artificial intelligence.
However, with the creation of a new human-like consciousness in the form of AI, humans are playing God and threatening to restart the evolutionary process in a new “species”. It is this ethical concern that is the most worrying when it comes to balancing the positives and negatives of AI creation and climate change.
Even if AI can be helpful in creating more accurate predictive models, is it worth creating an entirely new sentience that would only increase the complexity of managing an intelligent population on earth? Some may argue that AI will never really reach this level of human-level intelligence and phenomenology, but researchers are actively working towards this end. Therefore, those concerned about the climate and the future health of the planet should be equally as involved in the discussion around AI and its potential affects on climate.
This insight is a part of our Undergraduate Seminar Fellows’ Student Blog Series. Read work from other students and learn more about the Undergraduate Climate and Energy Seminar.