This video talks about how artificial intelligence (AI) can be used to reduce climate change impact, but also discusses the environmental cost of AI itself. AI can help us reduce climate change by improving efficiency, for example, by helping airplanes create fewer vapor trails and recommending eco-friendly routes. However, training large AI models requires a lot of energy, and the hardware used in this process can also be very energy-intensive. For example, a study showed that training a large language model called Bloom used more energy than the average American home for four decades. Additionally, AI models often require a lot of water to cool themselves down. Researchers estimate that a large language model called GPT-3 needs about 500 milliliters of water for a basic conversation. Companies can reduce the environmental impact of AI by using it only when necessary and by housing AI models in locations with cleaner energy sources.
