Artificial Intelligence (AI) is taking the world by storm, growing not only in the number of systems available, but also in complexity and capability. Fueling this expansion requires a monumental amount of energy to meet the increasing demand for computing power.
But at what environmental cost?
While it’s no secret that AI has a significant environmental footprint, especially considering the energy consumption of data centers that are constantly powered and require extensive cooling, determining its true carbon impact has been a challenge. The lack of a standardized method for measurement makes it difficult to quantify the exact impact of AI’s energy needs.
With AI expected to grow 30-40% annually over the next decade, it’s clear that this technology is here to stay and will only get bigger. “A few studies have estimated the carbon footprint of individual AI systems, such as GPT-3, especially after they became popular,” said Meng Zhang, a researcher at Zhejiang University in China. However, very few have attempted to calculate the combined emissions of the world’s major AI systems to understand their collective footprint.
“[If we have this figure]researchers and the public can better understand what AI is like [actually] affect the environment. It will also attract the attention of environmentalists, AI developers and policy makers, helping us to use AI in a more sustainable way.”
Place a number with AI
In their study, Zhang and his team calculated the carbon emissions released between 2020 and 2024 from 79 known AI systems, including Gemini Ultra, GPT-4, Mistral Large and Inflection-2.
Central to their estimates was the calculation of the energy consumption of graphics processing units (GPU) — a computer chip originally designed to render graphics, but now used in training AI systems, such as deep neural networks, where large amounts of data must be processed simultaneously. As such, GPUs are the main consumers of energy in these systems.
The team made estimates of carbon emissions based on single training runs and then extrapolated these to calculate the total emissions from both training and using the AI systems. What they found was that collectively the top 20 AI systems they included in their study consumed enough energy to compete with a small country, such as Iceland or the Democratic People’s Republic of Korea as examples. In fact, in 2022, the carbon emissions from these AI systems will surpass the emissions emitted by 137 individual countries.
The team also predicted that the projected total carbon footprint of the AI systems could reach up to 102.6 Mt CO2 equivalent per year — similar to the emissions of 22 million people over the course of a year.
While alarming, Zhang says these estimates may be just the tip of the iceberg, given the pace at which AI is expanding. For example, ChatGPT-3.5 has about 175 billion parameters – internal variables and values that the model uses to make predictions or generate outputs. The more parameters a model has, the more complex and powerful it is, but it also requires more computing power to train.
ChatGPT-4, on the other hand, can contain up to 1.8 trillion parameters and its emissions are estimated to be twelve times more than ChatGPT 3.5, according to Zhang.
Can we curb AI’s environmental impact?
It seems paradoxical to say we can use AI sustainable while continuing to grow so rapidly.
Although Zhang points out that renewable energy is a promising solution, it is not yet scalable enough to meet the rapidly increasing energy needs of a growing population, let alone AI and other related industries. Renewable energy infrastructure still faces challenges in terms of global adoption and efficiency.
Zhang and his team also framed their analysis in the context of carbon taxes, estimating that if applied to AI’s energy use, it could incur costs of around $10 billion USD per year.
“To get the attention of policy makers, we converted our carbon emission data into financial figures using carbon taxes,” he explained. “We believe this will really get their attention, especially given the urgent need for sustainable development [globally].”
This could potentially incentivize companies to reduce their carbon footprints by making polluting activities more expensive. However, the impact of carbon taxes will depend on their scale and how effectively they are implemented. For example, companies can be incentivized to use cleaner energy sources or make AI systems more energy efficient to avoid fines. But the effectiveness of carbon taxes could be limited if renewable energy sources remain insufficient or if companies can pass the cost on to consumers.
In addition, without global cooperation, there is a risk that emissions may simply shift to regions with laxer environmental regulations. Ultimately, while carbon taxes can help mitigate the environmental impact of AI, they are not a complete solution on their own and will need to be part of a broader set of policies and technological advances to comprehensively address the issue.
While these are pressing issues to consider, quantifying the impact of AI systems is an important first step. Zhang believes their method of calculation is a fair one, but says more work is needed.
“Given the potential financial impact of future policies, such as carbon taxes, we need even more precise methods,” he said. “This includes deciding on clear boundaries for what we calculate, such as whether to include the production of AI hardware, to understand exactly how much energy is used during the AI’s derivation process, and to estimate the carbon emissions of the type to calculate electricity used.”
A double-edged sword
More transparency is needed about how such energy needs and carbon emissions are calculated, and the current study’s calculations may only be an approximation.
For example, the team used a default value of carbon intensity of electricity production double that of the global average as estimated by Our world in data. Coupled with their assumption that ChatGPT growth rate is representative of the entire AI sector and that carbon emissions from distractions are 1000 times that of training, their predicted emissions may be overestimated.
It will take time to collect the correct data and may require cooperation from the companies that manage and develop such systems.
One thing is for sure: AI is here to stay, and that’s not necessarily a bad thing, given the benefits it brings to society. Like any technology, it has its pros and cons. For one, Zhang says, it can help in our efforts to become a more sustainable society.
“AI has the potential to significantly benefit the environment by enabling more efficient resource management, advancing renewable energy technologies and improving climate change modeling,” Zhang said. “It’s a double-edged sword.”
Reference: Yu, Y, et al., Revisiting the Environmental Impact of Artificial Intelligence: The Overlooked Source of Carbon Emissions? Front. Environment. Science. Eng (2024) DOI: 10.1007/s11783-024-1918-y
Feature Image Credit: Shubham Dhage on Unsplash