AI’s Growing Energy Demands
One of the forefront researchers, Sasha Luccioni, warns that generative AI is significantly more energy-hungry than traditional searches. Unlike conventional methods, AI generates new information, which requires a substantial amount of computational power.
Efforts have been made to quantify these emissions. Luccioni, recognized as one of the top 100 influential figures in AI, has dedicated years to measuring the carbon footprint of popular AI programs like ChatGPT.
With AI models requiring enormous computing capacities, the energy used for training and individual user requests is immense. The International Energy Agency reports that AI and cryptocurrency sectors consumed nearly 460 terawatt hours of electricity in 2022.
This consumption accounts for a staggering two percent of total global production, raising concerns about the sustainability of such technologies in the face of climate change.
Innovative Tools for Measuring Impact
Luccioni has been at the forefront of developing tools to measure AI’s environmental impact. In 2020, she helped create “CodeCarbon,” a tool downloaded over a million times, enabling developers to quantify the carbon footprint of their code.
Her current project involves creating a certification system for algorithms, similar to the US EPA’s energy scores for electronic devices. This system would provide energy efficiency ratings for AI models, encouraging better decision-making by users and developers.
Although this system doesn’t account for water or rare materials, it offers a way to measure energy efficiency. For example, an AI model could be rated A+ for efficiency, while another might receive a D.
Such transparency is crucial for promoting sustainable practices in AI development and usage, Luccioni argues.
The Push for Transparency
To refine her tools, Luccioni works with open-source generative AI models but faces resistance from commercial giants like Google and OpenAI. Despite commitments to carbon neutrality, these companies saw significant increases in emissions due to AI.
Google’s emissions rose by 48 percent compared to 2019, while Microsoft saw a 29 percent increase from 2020. This highlights the urgent need for transparency in AI energy consumption.
“We are accelerating the climate crisis,” Luccioni asserts, calling for more openness from tech companies. Governments, she says, are “flying blindly” without clear data on AI’s environmental impact.
She believes that once transparency is achieved, legislation can follow to mitigate AI’s environmental footprint.
Advocating for Energy Sobriety
It’s crucial to educate the public about what generative AI can and cannot do, and the associated costs. Luccioni’s latest study shows that producing a high-definition image with AI consumes as much energy as recharging a cell phone.
As companies increasingly integrate AI into daily life, with conversational bots and connected devices, the concept of energy sobriety becomes more important.
Luccioni emphasizes that the goal is not to oppose AI but to choose tools wisely and use them judiciously. This approach can help balance technological advancement with environmental sustainability.
By understanding AI’s energy demands, users and developers can make informed decisions that align with climate goals.
© 2024 AFP
Citation: AI is ‘accelerating the climate crisis,’ expert warns (2024, September 15) retrieved 15 September 2024 from https://techxplore.com/news/2024-09-ai-climate-crisis-expert.html
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without the written permission. The content is provided for information purposes only.
Luis
OMG, the energy use is insane! Can’t believe it matches entire nations!
oliver
This makes me think twice about using AI technologies. We need sustainable solutions ASAP.
pepperunity6
How does the energy consumption of AI compare to other industries?
emiliasolstice0
Thanks for the insights! We need more transparency from big tech companies.
Hannah_Infinity
Great article, but what about the impact on rare materials? Are there any studies on that?
DavidQuantum
So basically, using AI is like leaving a light bulb on 24/7? 😅
PiperEclipse0
Honestly, this is quite alarming. Are there any initiatives to regulate AI’s energy consumption?
EthanDelta
Is there any way to make AI more energy-efficient?
Ayden_Whispering
Wow, I had no idea AI could be so energy-intensive! Thanks for sharing this information. 👍