Artificial intelligence (AI) is a technology that requires significant energy resources for both training and use. The more advanced the AI system, the more energy it consumes. For instance, a single query on ChatGPT consumes three times more energy compared to a Google search.
OpenAI, a prominent company in the AI landscape, is known for creating the ChatGPT chatbot. To train the GPT-3 language model, an estimated 78,437 kWh of electricity is required. This highlights the massive energy consumption involved in AI training, which demands large amounts of data and computational power. As AI models become more complex, their energy requirements are likely to increase.
The energy consumption not only impacts electricity usage but also translates into significant water usage for cooling AI systems. Researchers have estimated that training GPT-3 in one of Microsoft’s advanced data centers, a partner of OpenAI, directly consumed 700,000 liters of clean fresh water – equivalent to producing 370 BMW cars or 320 Tesla electric vehicles. Google’s own data centers in the US used 12.7 billion liters of fresh water in 2021 for cooling processes.
In addition to water usage, AI technology consumes substantial amounts of water during operation. According to a study, a single conversation of 20 to 50 questions with ChatGPT requires about 500 ml of water for cooling. Furthermore, the energy consumption of AI systems is significantly higher compared to traditional processes. For example, a Google search consumes an average of 0.0003 kWh, while a query in ChatGPT requires between 0.001 and 0.01 kWh.
The environmental impact of AI’s energy and water consumption raises concerns about sustainability and resource management. As AI continues to advance, it is crucial to develop more energy-efficient and environmentally friendly alternatives to mitigate its impact on our planet.
