Microsoft has stated that it is working on ways to make large systems more efficient and will soon measure the energy and carbon footprint of its AI research. The company aims to be carbon negative, water positive, and zero waste by 2030. OpenAI has also acknowledged the energy and water-intensive nature of training large models and is working to improve efficiencies.
Key takeaways:
- Microsoft's global water usage increased by 34% in 2021-22, largely due to AI research and its partnership with OpenAI.
- ChatGPT, a large language model (LLM), consumes 16 ounces of water for every 5 to 50 user prompts due to the water-based liquid cooling in massive data centers.
- Despite environmental concerns, LLMs like ChatGPT have not faced the same level of scrutiny as energy-consuming practices like cryptocurrency mining.
- Microsoft plans to monitor its emissions and increase its use of clean energy to meet its sustainability goals of being carbon negative, water positive and zero waste by 2030.