Using AI has an environmental impact — here are 4 ways you can minimize it

Artificial intelligence has become increasingly popular in recent years, with data centers consuming 414 terawatt-hours of electricity annually, equivalent to about 1.5 percent of global electricity use. This demand is projected to double by 2030. The majority of this increased demand is met by fossil fuels, and there are concerns about the use of local water resources to cool data centers in drought-stricken regions.

While the environmental impact of AI is relatively small for an individual, tech companies and users can take steps to minimize it.

One simple way to reduce the environmental impact of AI is to carefully consider whether it is truly necessary for a given task. For example, asking a chatbot a simple question like "What should I wear today?" or "How is the weather?" is unnecessary and can waste energy. Turning off AI-generated responses in search engines, like Google's AI overviews or Bing's Copilot search, can also help save energy.

Additionally, selecting "Web results only" in a browser or including "-ai" in a web search query can further reduce energy consumption.

Another approach is to shift to smaller language models that are specialized for specific tasks, such as translation or summarization. These models are trained more narrowly and perform fewer computations, resulting in less energy consumption compared to large, all-purpose language models. For instance, a study by UNESCO found that using smaller models, such as opus-mt-en-es for English-Spanish translations, consumed between 15 and 20 times less energy than the more popular Llama 3.1 model.

Although smaller models may not be as user-friendly, they are freely available from AI platforms like Hugging Face.