Climate Activist Groups Advocate for Legislation on Carbon Emissions from AI Systems
In an effort to combat climate change, climate activist groups are urging the US government to impose regulations on tech companies that require them to disclose the carbon emissions generated by their artificial intelligence (AI) systems. A signed letter, addressed to Senate majority leader Chuck Schumer, includes the endorsements of 22 organizations, including Amazon Employees for Climate Justice, Fight for the Future, Green America, the Union of Concerned Scientists, Friends of the Earth, and Accountable Tech.
Schumer, recognizing the importance of regulating AI, has been actively seeking to demystify the technology and its implications. With legislation in the works, he aims to ensure that AI is taken as seriously as national security. The activist groups agree that AI has the potential to benefit society, but they express concerns about the environmental impact of widespread usage of large language models (LLMs), which consume a significant amount of energy and contribute to carbon emissions.
The letter outlines several policy demands, including public reporting of energy use and emissions throughout the entire life cycle of AI models, encompassing training, updates, and search queries. Furthermore, the activist groups call for accountability among companies and their executives for any environmental or other harm resulting from the use of generative AI.
Underreported Carbon Emissions by Big Tech
Technology giants, such as Google, IBM, Microsoft, and SAP, have been accused of underreporting their carbon emissions even before the current AI boom. Google claims to have addressed emissions from the development of AI models and asserts to have achieved zero emissions in 2020, though inconsistencies have been observed in its reporting. Some companies divulge only half of their emissions. IBM recently launched a cloud carbon calculator tool to identify patterns and anomalies associated with higher greenhouse gas emissions.
Big Tech’s extensive energy consumption is primarily driven by data centers, which require substantial amounts of energy and water for cooling purposes. Google’s data centers alone consumed twice as much energy as San Francisco in 2020, and water consumption increased by 20% last year.
The Carbon-Intensive Nature of AI Training
Training a single AI model or chatbot can consume more electricity than 100 US homes in a year. The energy used during training for ChatGPT, for instance, amounted to 1.287 gigawatt-hours, equivalent to the annual consumption of 120 American homes. Creating GPT-3, with its 175 billion parameters, was found to be as energy-intensive as driving 123 gasoline-powered vehicles for a year. Data centers globally emit more carbon than the commercial airline industry, and in the US, they account for 2% of all electricity usage, according to the Department of Energy. Experts predict that computing technology could consume 8% to 21% of global energy by 2030, with data centers contributing to a third of that consumption.
Combatting Climate Disinformation with AI
The climate activist groups not only raise concerns about the energy consumption of LLMs like ChatGPT, but they also highlight the potential for AI models to amplify climate change disinformation. As these models can rapidly generate and distribute misleading information, there is a risk of perpetuating climate denialism and impeding efforts to combat climate change. The groups argue that monitoring and transparent disclosure of these AI technologies are crucial for consumers and policymakers to understand the trade-offs involved.
Denial of responsibility! Vigour Times is an automatic aggregator of Global media. In each content, the hyperlink to the primary source is specified. All trademarks belong to their rightful owners, and all materials to their authors. For any complaint, please reach us at – [email protected]. We will take necessary action within 24 hours.
