Nowadays, AIās applications extend into countless industries and to nearly every single facet of our lives. It is tackling the worldās greatest challenges, including the climate crisis. AIāwith its immense capabilitiesāseems like a silver bullet solution to climate change. However, it poses hidden, yet significant environmental concerns that require further scrutiny.
In this edition of Hold the Code, weāll take a deeper dive into AIās role in fightingāand contributingāto the climate crisis. Itās part of a larger series that examines AIās role in a specific problem in depth.
Lastly, be sure to check out HTCās crossword scavenger hunt and win free Tomate! The final clue is our Executive Ianās favorite vegetable, which is broccoli! The rest of the answers are on our page. As always, happy reading and be sure to share HTC with a friend!
AI Fights Climate Change
When people first think of AI, they may think of robots or deep fakes ā the environment and climate activism is rarely the first association. But artificial intelligence can be an impressive tool to bolster efforts across climate research, finance, education, and modeling.Ā
AI has the potential to fight climate change on an unprecedented scale, in the following ways:Ā
AI and Self-Driving Cars: AI self-driving cars may reduce emissions by 50 percent by 2050 by identifying the most efficient routes.Ā
AI and Agriculture: Peanut farmers in India increased their harvest by 30 percent by using AI technology. This ādigital agricultureā involves the farmers receiving automated voice calls or text messages telling them when to sow the seeds and when weather or pests might disrupt their crops.Ā
AI and Natural Disasters: Using AI, scientists can identify patterns in data sets and make weather predictions. This technology can reduce the devastation from storms, wildfires and droughts, and warm people who live in at-risk areas to evacuate. For example, when researchers provide AI with seismic and geographical information, the technology can accurately predict future volcanic eruptions.Ā
The intersection between climate change and machine learning has been investigated extensively by the group Climate Change AI. In a presentation called āClimate Change 101,ā Climate Change AI identified five more ways that machine learning can test climate change solutions.Ā
Forecasting extreme events
Approximating time-intensive simulations regarding climate and energyĀ
Accelerated experimentation involving batteries, perovskites, nuclear fusionĀ
Optimizing systems of agriculture, heating and cooling
Predictive maintenance for natural gas pipelines, creating resilient infrastructureĀ
Machine learning cannot replace humans, but it can help humans make better climate decisions by simulating the outcomes of events and checking projects. Using AI, we can create a more nuanced picture of our future and identify fail-safe solutions.
AI Contributes to Climate Change
Even with greater algorithm efficiency, Artificial Intelligence usage requires large amounts of electricity and data processing, which all contribute to fossil fuel emissions. Of particular concern is the training models/data processing that often require millions of distinct data points. According to a study from researchers at UMass Amherst, training a single natural-language processing AI model can emit around 626,000 pounds of carbon dioxide, equivalent to 5 times the lifetime emissions of the average American car, or 315 round-trip flights between San Francisco and New York City.
As AI developers seek to create more comprehensive and accurate tools, they will require larger sets of data to train models, which ultimately leads to greater energy consumption. Given the economic advantages AI tools can provide a company if successfully developed, there exist strong incentives to ignore the long-term environmental impacts of AI in favor of optimizing for short-term profits. As explained by computer scientist and researcher Carlos GomĆ©z-RodrĆguez in an article for the MIT Technology Review, āmuch of the latest research in AI neglects efficiency, as very large neural networks have been found to be useful for a variety of tasks, and companies and institutions that have abundant access to computational resources can leverage this to obtain a competitive advantage.ā
Of equal significance is the storage and transmission of data used within AI training programs, whichĀ āconsume about 200 terawatt hours of power per year.ā This trend appears to be moving upward, according to researchers Anders Andrae and Tomas Edler:Ā
ācomputing and communications technology will consume between 8 percent and 20 percent of the worldās electricity, with data centers accounting for a third of that.ā
What do we do?
Despite the potential AI has to combat climate change, itās clear that it has serious environmental issues. So, what do we do from here?
Some companies have taken measures into their own hands. For example, data farms, like some in Iceland, have shifted to running entirely on clean energy. Powered by its hydroelectric and geothermal resources, Iceland has become a popular location for new data centers. Rather than relying on energy-intensive fans or air conditioning, these data centers take advantage of Icelandās cold climate. However, Icelandās climate is unique, and most countries arenāt able to replicate its environmental conditions.Ā Ā
Several other potential methods of reducing AIās carbon footprint include:
AI researchers can track the carbon footprint of their algorithms through websites and other tools
Code that tracks the energy usage of individual computer chips within the program itself
Public disclosure of carbon emissions estimates as a result of running these algorithms
Reporting on computational costs of training algorithms to enhance transparency
However, none of these methods are implemented on a wide scale, and there is little consistency in which they are enforced. Greater regulation from a policy standpoint is needed. Regulation could take the form of prohibiting the sale of less efficient GPU chips or offering incentives for companies to use more efficient chips.Ā
If we are serious about the climate crisis, we must look inwards and critically access the tools weāre using to promote environmental sustainability.Ā
Love HTC?ā¤ļø
Follow RAISO (our parent org) on social media for more updates, discussions, and events!
Instagram: @Raisogram
Twitter: @raisotweets
RAISO Website: https://www.raiso.org