AI’s energy consumption poses environmental problems

Training an advanced AI model takes time, money, and high-quality data. It also takes energy – lots of energy.

Between storing data in large-scale data centers and using that data to train a machine learning or deep learning model, the power consumption of AI is high. While an AI system can be financially profitable, AI poses an environmental problem.

AI energy consumption during training

Take some of the most popular language models, for example.

OpenAI trained its GPT-3 model on 45 terabytes of data. To train the final version of MegatronLM, a language model similar to but smaller than GPT-3, Nvidia ran 512 V100 GPUs over nine days.

A single V100 GPU can consume between 250 and 300 watts. If we assume 250 watts, then 512 V100 GPUs consume 128,000 watts, or 128 kilowatts (kW). Running for nine days means that training the MegatronLM cost 27,648 kilowatt hours (kWh).

The average household uses 10,649 kWh per year, according to the US Energy Information Administration. Therefore, the formation of the final version of MegatronLM used almost the amount of energy used by three households in one year.

New training techniques are reducing the amount of data needed to train machine learning and deep learning models, but many models still need huge amounts of data to complete an initial training phase, and to additional data to keep up to date.

Data Center Power Consumption

As AI becomes more complex, expect some models to use even more data. This is a problem, because data centers use an incredible amount of energy.

“Data centers are going to be one of the most impactful things on the environment,” said Alan Pelz-Sharpe, founder of analytics firm Deep Analysis.

AI has many benefits for businesses, but it creates problems for the environment.

IBM’s Weather Company processes about 400 terabytes of data per day to enable its models to predict weather days in advance around the world. Facebook generates about 4 petabytes (4,000 terabytes) of data per day.

People generated 64.2 zettabytes of data in 2020. That’s about 58,389,559,853 terabytes, market research firm IDC estimated.

Data centers store this data around the world.

Meanwhile, the largest data centers require more than 100 megawatts of electrical capacity, which is enough to power some 80,000 US homes, according to energy and climate think tank Energy Innovation.

With around 600 hyperscale data centers — data centers that exceed 5,000 servers and 10,000 square feet — around the world, it’s unclear how much energy is needed to store all of our data, but the number is likely staggering. .

From an environmental point of view, the energy consumption of data centers and AI is also a nightmare.

Google data center
A Google data center in Douglas County, Georgia.

AI, data and environment

The use of energy creates CO2, the main greenhouse gas emitted by humans. In the atmosphere, greenhouse gases like CO2 trap heat near the Earth’s surface, causing the Earth’s temperature to rise and throwing delicate ecosystems out of balance.

“We have a crisis in energy consumption,” said Gerry McGovern, author of the book. Waste from around the world.

AI is energy-intensive, and the higher the demand for AI, the more energy we use, he said.

“It’s not just electrical energy to train an AI,” he said. “He builds the supercomputers. He collects and stores the data.”

McGovern pointed to estimates that by 2035, humans will have produced over 2,000 zettabytes of data.

Data centers are going to be one of the most impactful things on the environment.

Alan Pelz-SharpeFounder, In-Depth Analytics

“The storage energy needed for this will be astronomical,” he said.

Currently, the biggest data users are doing little to address the carbon footprint or energy consumption of AI.

“I am aware of a certain recognition [of AI’s carbon footprint problem] but not much action,” McGovern said. “Data centers, which are the ‘food source’ for AI, have focused on electrical efficiency and have certainly made major improvements over the past 10 last years.”

While data centers have become more electrically efficient over the past decade, experts estimate that electricity only accounts for about 10% of a data center’s CO2 emissions, McGovern said. The infrastructure of a data center, including the building and the cooling systems, also produces a lot of CO2.

In addition to this, data centers also use a lot of water as a form of evaporative cooling. This method of cooling reduces electricity consumption, but can consume millions of gallons of water per day per large-scale data center. Additionally, the water used can be polluted in the process, McGovern noted.

“There’s always this blanket assumption that digital is inherently green, and it’s far from it,” he said.

Environmental impact of companies

While the average business can’t change the way large companies store their data, businesses that care about their environmental footprint can focus on creating high-quality data rather than massive amounts of it. They can delete data they no longer use, for example; companies tend not to use 90% of data 90 days after storing it, according to McGovern.

Companies can also adjust their use of AI or the type of AI they use.

Organizations can think about the specific use case they want to accomplish and choose an AI or automation technology dedicated to that use case. However, different types of AI have additional power consumption costs.

Companies can get carried away with the idea that they need an advanced deep learning system that can do it all, Pelz-Sharpe said. However, if they want to tackle a targeted use case, like automating an invoicing process, they don’t need an advanced system. These systems are expensive and data-intensive, which means they have a high carbon footprint.

A dedicated system will have been trained on a much smaller amount of data while probably complementing a specific use case as well as a more general system.

“Because it is highly specialized, this AI was trained on the most accurate data possible” while maintaining a small data set, Pelz-Sharpe said. A deep learning model, on the other hand, has to produce massive amounts of data to accomplish anything.

“In all of our decisions, we have to consider Earth experience,” McGovern said.

Comments are closed.