AI Climate Promises Are Often Greenwashing
Tech giants engage in greenwashing when they claim that Generative AI can help combat climate change. This is shown by an analysis of 154 statements from leading companies and organizations. The study, commissioned by nonprofits such as Beyond Fossil Fuels and Climate Action Against Disinformation, found not a single example in which popular tools like Google’s Gemini or Microsoft’s Copilot led to a “material, verifiable and substantial” reduction in greenhouse gas emissions, reports the Guardian.
Tried-and-Tested Greenwashing Methods
Most climate protection claims relate to machine learning, not the energy-intensive chatbots and image generation tools that drive the explosive growth of power-hungry data centers. Many corporations falsely conflate their traditional AI with Generative AI in their claims.
Energy analyst Ketan Joshi, author of the report, describes the industry’s tactics as “distracting” and compares them to tried-and-tested greenwashing methods. He draws parallels to fossil fuel companies that promote modest investments in solar panels and overstate the potential of carbon capture. “These technologies avoid only a tiny fraction of emissions relative to the massive emissions from their core business,” Joshi explains. “Big Tech has adopted this approach, upgraded it, and expanded it.”
LLMs and GenAI Harmful to the Planet
Only 26 percent of the examined climate claims cited published academic research, while 36 percent provided no evidence at all. For example, there is a widespread claim that AI could help reduce global greenhouse gas emissions by 5 to 10 percent by 2030. Google repeated this figure as recently as April of last year, but it came from a report that Google itself had commissioned.
Sasha Luccioni, AI and Climate Lead at Hugging Face, comments: “When we talk about AI that is relatively bad for the planet, it’s mainly generative AI and large language models. When we talk about AI that is ‘good’ for the planet, it’s often predictive models, extractive models, or old-school AI models.”
Data Center Power Consumption Rising
Data centers currently consume only one percent of global electricity, but their share of US power consumption is reported to rise to 8.6 percent by 2035, representing more than a doubling. The International Energy Agency forecasts that data centers will account for at least 20 percent of electricity demand growth in industrialized countries by the end of the decade.
While a simple text query to ChatGPT consumes about as much energy as a light bulb for one minute, consumption increases significantly for complex functions such as video generation and deep research.
A Google spokesperson defends the company’s methodology; Microsoft declined to comment. The report was released during the AI Impact Summit in Delhi this week.
Joshi calls for the discussion about AI’s climate benefits to “return to reality. Falsely coupling a large problem with a small solution serves as a distraction from the very avoidable damage caused by unchecked data center expansion,” warns the analyst.
