Last month, the San Francisco Chronicle published an op-ed by me about the environmental impact of AI tools now being integrated into many areas of our lives, and used (for different purposes and with various levels of choice) by millions of people.
The op-ed joined a growing chorus of voices calling for more awareness of this topic, more transparency from AI purveyors, and more research into the development of “green AI.” I’ve listed eight such articles below—the earliest one being from 2019, published in the MIT Tech Review.
After the Chronicle piece came out, I was still startled by the number of people who reached out to tell me that was the first they’d heard about the issue of AI’s environmental impact. We need many more articles (and research!) on this, to help inform any organizations considering the integration of AI into their products and services, and any users who are taking advantage of the opportunity to experiment with various AI tools.
The following is not a comprehensive reading list, but it offers articles that add key details. I hope you will share it with others who might be interested in learning more.
- “Training a Single AI Model Can Emit as Much Carbon as Five Cars in Their Lifetimes: https://www.technologyreview.com/2019/06/06/239031/training-a-single-ai-model-can-emit-as-much-carbon-as-five-cars-in-their-lifetimes/ (from 2019, long before the current boom in generative models)
- “The Mounting Human and Environmental Costs of Generative AI”: https://arstechnica.com/gadgets/2023/04/generative-ai-is-cool-but-lets-not-forget-its-human-and-environmental-costs/ (linking to Dr. Sasha Luccioni's research paper comparing the impact of multiple models: https://arxiv.org/abs/2211.02001)
- “The Secret Water Footprint of AI Technology”: https://themarkup.org/hello-world/2023/04/15/the-secret-water-footprint-of-ai-technology
- “AI Can Help in the Fight Against Climate Change, But If It’s Not Deployed Wisely, Artificial Intelligence Could Mess Things Up More”: https://www.businessinsider.com/ai-can-help-climate-change-2023-5
- “AI Chatbots Lose Money Every Time You Use Them. That Is A Problem.” https://www.washingtonpost.com/technology/2023/06/05/chatgpt-hidden-cost-gpu-compute/
- “How to Make Generative AI Greener”: https://hbr.org/2023/07/how-to-make-generative-ai-greener
- “Silicon Valley’s New Toy Has a Risky Tradeoff”: https://slate.com/technology/2023/08/chatgpt-ai-arms-race-sustainability.html (linking to a report released by OpenAI in 2018: https://openai.com/research/ai-and-compute)
In “How to Make Generative AI Greener,” Ajay Kumar and Tom Davenport include a suggestion that warrants reiteration:
Be discerning about when you use generative AI. Machine learning and NLP tools are revolutionary for medical-related health problems and prediction. They are great for predicting natural hazards such as tsunamis, earthquakes, etc. These are useful applications, but… generating blog posts or creating amusing stories may not be the best use for these computation-heavy tools. They may be depleting the earth’s health more than they are helping its people.
Of course, AI is also being used in efforts to combat climate change and reduce the degradation of the environment. Being discerning about AI usage requires a careful analysis of the benefits and harms of different uses—and of the stakeholders to whom the benefits and/or harms accrue.
As I pointed out in my op-ed, “[e]ven discussions specifically devoted to ethical concerns related to #AI (highlighting issues like bias, privacy violations, indiscriminate scraping to build training data sets, etc.) often fail to include the environmental issues.” Hopefully, that will change soon.
Illustration: David Man & Tristan Ferne / Better Images of AI / Trees / CC-BY 4.0