Breaking Down the Carbon Footprint of Generative AI

While it is difficult to estimate the exact energy cost of a single AI model, the carbon footprint of these models is growing substantially.

Generative artificial intelligence (AI) refers to the ability of an algorithm to produce complex data, such as a sentence, a paragraph, an image, or even a short video.

Generative AI has long been used in applications like smart speakers to generate audio responses or in autocomplete to suggest a search query. But it only recently gained the ability to generate humanlike language and realistic photos.

However, these models are not cheap to train.

According to a report by the MIT Technology Review, training just one AI model can emit more than 626,00 pounds of carbon dioxide equivalent — which is nearly five times the lifetime emissions of an average American car.

What is Carbon Footprint of Generative AI Models?

It comes as no surprise that larger AI models require a ton of energy.

For example, in 2019, creating a generative AI model called BERT with 110 million parameters consumed the energy of a round-trip transcontinental flight for one person, according to Kate Saenko, an AI research scientist at FAIR Labs.

The much larger GPT-3, which has 175 billion parameters, consumed 1,287 megawatt hours of electricity and generated 552 tons of carbon dioxide equivalent just to get the model ready to launch.

As an AI researcher, I often worry about the energy costs of building artificial intelligence models. The more powerful the AI, the more energy it takes. What does the emergence of powerful generative AI models mean for society’s future carbon footprint?https://t.co/qXL9RGT3bv

— Kate Saenko 🤖♻️ (@kate_saenko_) May 24, 2023

The BLOOM model, developed by the BigScience project in France, is similar in size to GPT-3 but has a much lower carbon footprint.

A study by Google also found that using a more efficient model architecture and processor and a greener data center can reduce the carbon footprint by 100 to 1,000 times.

Generative AI Queries Produce More Carbon Footprint

The carbon footprint of a single generative AI query is estimated to be four to five times higher than that of a search engine query.

As chatbots and image generators become more popular, and as AI language models are incorporated into search engines, the number of queries they receive each day could grow exponentially.

One example of a popular chatbot built on generative AI is ChatGPT, which had over 1.5 billion visits in March 2023.

Microsoft has also incorporated ChatGPT into its search engine, Bing, and made it available to everyone.

While the energy costs of deploying AI assistants for search and other uses could add up, one upside is that asking a chatbot can be a more direct way to get information than using a search engine.

To make generative AI more efficient, more research is needed.

“The good news is that AI can run on renewable energy,” Saenko noted, adding:

“By bringing the computation to where green energy is more abundant, or scheduling computation for times of day when renewable energy is more available, emissions can be reduced by a factor of 30 to 40, compared to using a grid dominated by fossil fuels.”

Read More:

• Best AI Stocks to Invest in 2023
• Best AI Crypto Tokens & Projects to Invest in 2023
• 16 Best Crypto Under $1 for 2023 – Micro Caps Reviewed