Is AI Sustainable?
Through both its environmental and social impact, the development of artificial intelligence technologies poses both significant risks and opportunities for solutions.
AI & The Environment: Destruction or Deliverance?
As artificial intelligence continues to require increasing amounts of data, computing power, and energy, its environmental impact becomes an ever-important cause for concern. Rene Haas, Chief Executive Officer of chip design company ARM, stated that AI models like ChatGPT “are just insatiable in terms of their thirst for electricity,” and the data reflect this: The International Energy Agency (IEA) reports that a single request to ChatGPT requires enough electricity to power a 60-watt lightbulb for nearly three minutes and uses almost 10 times the power required by the average Google search. Today, AI data centers use approximately 4% of the United States’s electricity requirements. However, if current technological development trends continue without an increase in electrical efficiency, that number could reach up to 25% by 2030. As concerns about global climate change abound, the need for environmental consciousness is difficult to understate. Despite AI’s not-so-green energy needs, however, it also has the potential to offer a solution. The IEA estimates that roughly 10% of energy usage world-wide is wasted on inefficiently heating or cooling homes, office spaces, and other buildings. Artificial intelligence can be – and has been – used to optimize this process. Peter Herweck, Chief Executive Officer of sustainability-focused energy company Schneider Electric, predicts that the usage of AI to streamline these processes could reduce building energy consumption up to 25% by 2028. It is currently unclear whether these benefits will be enough to offset AI’s environmental harms. Nevertheless, despite not being as flashy as new, sustainable infrastructure, energy reduction is no small feat. Over the last 15 years, reduction in global energy consumption has saved the equivalent of 90 million barrels of oil per day, while solar energy has saved only 15 million per day in comparison.
Takeaway: When evaluating the impacts of AI as a whole, environmental repercussions must be a part of the conversation. Unless artificial intelligence can at least “break even” by saving as much energy as it consumes, we will all pay a price far greater than a subscription to ChatGPT plus. AI has the potential to solve the problem it has created, but only if it is used carefully and effectively. –Grace Ogden
To read more about AI’s environmental harms, please visit this article from The Wall Street Journal. To read more about AI’s energy saving potential, please visit this article from The Wall Street Journal.
Hungryroot
Due to excess production, packaging, storage, and delivery, food waste is an enormous contributor to climate change, both in the United States and around the world. Currently, approximately 33% of US food winds up in landfills, and the scientific journal Our World in Data found that food production accounts for 26% of greenhouse emissions globally through the food supply chain, livestock and fisheries, crop production, and land use. In an effort to combat our food waste problem, New York-based food subscription startup Hungryroot uses artificial intelligence to take data about customers’ preferences, allergies, health goals, budgets, and cooking habits to tailor recipes, portions, and grocery lists to each individual’s needs. Thus, the groceries purchased can last the week with no excess. Hungryroot is also able to reduce its own waste by curating recommendations based upon what their warehouses have in stock. The company has reported that these mindful, AI-powered practices have resulted in 80% less food waste than is produced by traditional grocery stores. When food waste is reduced, food production and the resulting greenhouse gas emissions can be reduced as well.
Takeaway: The impact of many driving factors of climate change can be mitigated simply by allocating resources more efficiently, and artificial intelligence is a powerful tool well-suited to tackling this problem. As climate change becomes an ever more pressing issue, we must take advantage of this technology to reduce waste wherever possible. –Grace Ogden
To read more about Hungryroot, please visit CNBC. To learn more about food waste statistics, please visit Our World in Data.
AI and Struggles with Realism
Artificial intelligence has been revolutionary in every industry over the past year. With increased use of AI, however, comes increased energy consumption. A research paper published in 2020, prior to the release of OpenAI’s ChatGPT that sparked public awareness of AI, established the potential for the development of AI to have a significant impact on carbon emissions. Major investments have been made into the development of AI by Meta, OpenAI, Microsoft, Google Ventures, Wells Fargo, and more, in direct contradiction to climate change policies established by some of those same companies. Microsoft’s carbon emissions for 2023 are up 30% yearly, in direct contradiction to their pledge to go carbon neutral by 2030 and coinciding with massive investments into the development of AI. The estimated carbon cost of training new models is huge, and more frightening, has yet to be accurately quantified across the board. With a lack of regulation and no accountability for companies to follow through on sustainable practices in powering AI development, the future of sustainable energy consumption in the nascent AI industry looks bleak. The double-edged sword of AI presents potential help in reducing energy waste while consuming energy at an unprecedented rate. Alongside sustainability concerns, generative AI models pose ethical concerns that are not being evaluated or legislated for at a rate appropriate to the development of the technology. The use of AI in scams, misinformation, fraud, and other illegal activities is exponential across the globe with limited tools available to identify, much less prevent, malicious AI use.
Takeaway: The social ramifications of unregulated AI use in cases like revenge porn, the political ramifications of election fraud and deep fakes, and the environmental crisis AI development is rapidly growing to be a major contributor towards all point to the need for legislation safeguarding the development of AI sooner, rather than later. - Natalie Sherman.
To read more about Microsoft’s pursuit of AI and its carbon emissions, please visit Bloomberg News. Read more about about AI legislation at the Hill; read about AI deep fakes in schools at Politico; read about deep fakes of Joe Biden at the Hill.