Environmental groups caution against claims AI solves climate crisis
Environmental groups caution against AI claims, warning of increased energy use and climate misinformation. AI advances, praised by big tech and the UN, are seen as tools to address global heating, tracking deforestation, pollution leaks, and extreme weather. AI is also used to predict droughts in Africa and measure melting icebergs.
Google, with its AI program Bard (now Gemini) and an AI project for more efficient traffic lights, leads in promoting emissions reductions through AI adoption. A recent report suggested AI could reduce global emissions by up to 10%, matching the EU’s carbon output by 2030. Kate Brandt, Google’s chief sustainability officer, emphasized AI’s significant role in addressing climate change, noting it’s at an “inflection point” for major environmental progress.
However, green groups’ new report questions the positive impact of the AI revolution on the climate crisis. It warns that AI will increase energy use from data centers and spread misinformation about climate science.
“We often hear that AI can save the planet, but we shouldn’t believe this hype,” said Michael Khoo, climate disinformation program director at Friends of the Earth, a member of the Climate Action against Disinformation coalition that published the report.
“AI won’t replace the internal combustion engine. People will be shocked by the increased energy consumption from AI and the spread of climate change misinformation.”
The new report highlights that the growing electricity needs of AI will lead to an 80% increase in planet-warming emissions if the number of data centers doubles, even with efforts to enhance their energy efficiency.
In the US, coal-fired power plants are being kept operational longer to meet the rising energy demands of AI. By 2027, AI servers could consume as much energy as Sweden, according to separate research.
This increased energy demand is largely due to the greater complexity of AI operations. Generating AI queries could require up to 10 times the computing power of a regular online search. Training systems like ChatGPT from OpenAI could consume as much energy as 120 US households in a year, the report suggests.
“There is no evidence to suggest AI will reduce energy use; all signs point to a significant increase due to new data centers,” said Khoo. “While there may be minor efficiency gains in data centers, the overall carbon emissions will rise.”
AI will also impede efforts to address the climate emergency by providing a platform for disseminating false or misleading information about climate science and global warming impacts, according to the coalition’s report.
This exacerbates an existing issue where major social media platforms, like Twitter/X, have become hubs for climate science denial, according to critics of AI.
“We’re witnessing AI fragmenting the information ecosystem at a time when we need unity,” Khoo explained. “AI is ideal for saturating the information space with quickly and inexpensively produced content of low quality. It’s clear how it could be used for climate disinformation. We’ll witness a relentless stream of micro-targeted climate misinformation content.”
The report emphasizes the need for improved transparency regarding AI’s energy consumption and for safeguards to monitor the spread of climate misinformation.
Jesse Dodge, senior research scientist at the Allen Institute for AI, expressed concern that AI could “accelerate” climate misinformation, potentially through methods like deepfake videos and images, while also increasing carbon emissions due to higher energy consumption.
“While this is a valid concern,” he explained, “the impact of AI applications themselves could potentially outweigh their electricity or water consumption. AI acts as an accelerant, allowing tasks to be completed faster. For instance, it could speed up oil extraction, but conversely, we have six teams using AI to mitigate the harm of climate change.”
One team focuses on climate modeling, another on tracking illegal fishing and endangered species, and there’s also monitoring and predicting wildfires. We believe there’s a net benefit overall, with a worthwhile trade-off.
Dodge expressed cautious optimism about AI’s largely positive impact on the climate crisis but emphasized the need for companies to be fully transparent and open about their energy use.
