TL;DR:
- The generative AI industry requires a significant amount of computational power and results in significant carbon emissions, the mining of rare minerals, and water and noise pollution.
- The crypto market is also having a disproportionate impact on the environment, with Bitcoin alone consuming 0.5% of the world’s electricity. AI language models and data centers are responsible for 1% of global carbon emissions.
- The generative AI industry prioritizes growth over efficiency, leading to an increase in environmental costs, which is expected to reach 3.2% of global carbon emissions by 2025 and potentially 14% by 2040.
- Despite being reported on for years, international governments have done little to address the environmental impact of AI.
- The pollution of AI is even more dangerous than the crypto industry as AI has produced a widely adopted product, and the measurement of pollution in its field is suppressed.
- The world is racing to reduce greenhouse gas emissions, and the AI industry must be part of this effort, with tech companies held accountable for the indirect environmental impacts caused by their software services.
Main AI News:
Generative AI is capturing the attention of investors and making headlines, but the cost of this tech race is often hidden from public view. A recent report by PitchBook sheds light on the environmental impact of the industry, as it requires a tremendous amount of computational power, leading to significant carbon emissions, mining of rare minerals, and water and noise pollution. In 2018, generative AI companies received $408 million in VC funding over 41 transactions. However, by 2022, that number skyrocketed to $4.5 billion over 110 transactions, while investment in crypto dropped by 70% from its peak.
According to PitchBook’s report, by 2025, 3.2% of all carbon emissions in the world will come from server farms, and the environmental cost is only expected to increase as the industry prioritizes growth over efficiency. By 2040, server farms may account for as much as 14% of global carbon emissions, making it even more challenging to combat the climate crisis. The need for more server farms is driving the industry to underpopulated areas, where low-cost electricity, multiple water sources, cheap land, and relaxed planning regulations are available.
In Israel, for example, the electricity consumption of the ten planned server farms is equivalent to 250,000 households or an equivalent of 170,000 cars on the road, according to estimates by consulting firm BDO. However, the country’s relatively low electricity prices, roughly half that of most European countries, make it an attractive location for larger corporations, especially after the recent increase in energy prices following the Russian invasion of Ukraine.
The Generative AI Industry: A Costly Pursuit
The generative AI industry, like the crypto market, is pushing forward with efforts to replace existing systems, regardless of the environmental cost. The crypto market, which primarily serves as an alternative trading arena, is already consuming massive amounts of energy long before it comes close to replacing traditional financial markets. According to the Cambridge Bitcoin Electricity Consumption Index (CBECI), Bitcoin alone consumes 0.5% of the world’s electricity, and if it were a country, it would rank 32nd in terms of electricity consumption.
Similarly, the generative AI sector is already having a significant impact on the environment, and this is just the start. The International Energy Agency reports that AI language models and data centers are responsible for 1% of global carbon emissions. ChatGPT, which serves approximately 13 million users daily and enables Bing to handle half a billion queries per day, consumed as much electricity as 175,000 people in January. An analysis by the University of Berkeley suggests that the training of GPT-3, the basis of the first version of the chat, resulted in the emission of 550 tons of carbon dioxide.
The crypto market involves a substantial investment in server farms, also known as “mining farms,” which power the blockchain technology. Historically, companies in this field have funded their operations through the sale of minted coins, but as the crypto market continues to grow, they have begun to raise capital and debt from venture capital funds and financial institutions like Goldman Sachs, Fidelity, and JP Morgan, which now offer crypto trading and management services.
The cost of this industry extends far beyond just the massive electrical consumption, and also includes water and noise pollution. The environmental impact of the crypto market is not limited to just high electricity consumption, but it also includes water and noise pollution. An environmental working group has conducted a study to examine the type of pollution and its direct impact on communities living near mining farms.
The Bitcoin mining company Greenidge in Dresden, New York, draws a large amount of water from nearby Seneca Lake to mine Bitcoin, which is returned at a high temperature that is harmful to local fish and humans. Another mining center in Adel, Georgia, produces noise 24/7, affecting nearby communities and causing ripples in nearby lakes. The generative AI industry, like the crypto market, is pursuing growth at a high cost to the environment and must be evaluated carefully as these industries continue to expand.
The Polluting Reality of Generative AI: A Critical Assessment
Generative AI is not only a polluting industry but also an industry whose growth is inherently polluted. For increased performance, an exponentially larger model is required, which leads to an increase in training data and computational costs and, therefore, carbon emissions.
The prevailing belief in the field is that the bigger the model, the better, with only a few models showing a decrease in training costs, as in the case of Facebook’s latest model. A 2019 study by the University of Massachusetts Amherst found that the carbon footprint of training one large language model was 284,000 tons of carbon dioxide over its lifetime.
The pollution of AI is even more dangerous than the crypto industry in some ways. Unlike crypto, AI has produced a product that has been widely adopted on a large scale by the general public, and the measurement of pollution in its field is not part of the discourse and is actively being suppressed.
OpenAI was one of the pioneers in reporting and measuring pollution in AI. In 2018, OpenAI published a widely cited report that stated the surge in computing power for various AI training doubled every 3.4 months or by a factor of ten every year. Five years later, OpenAI released its latest and most powerful model, GPT-4, while deliberately hiding the computational power required to train and deploy it.
The world is racing to reduce greenhouse gas emissions across various industries to curb the worst effects of climate change, which the UN has called “the defining issue of our time.” The current decade is critical in limiting the increase in temperature, and greenhouse gas emissions must be reduced by approximately 50% to meet international climate commitments. The pollution of the generative AI industry must be addressed as a critical part of this effort.
AI’s Contribution to the Climate Crisis: A Call for Action
Tech companies often boast about their efforts to fight the climate crisis, but in reality, many are increasing their emissions in the current race to perfect AI capabilities. Companies like Google, Microsoft, and Amazon have set goals to reach net-zero greenhouse gas emissions in the coming decades and participated in the global climate strike in 2019 to raise awareness of the growing dangers of the climate crisis.
However, the need for action to recognize and reduce the pollution caused by the AI industry is becoming more pressing. While many other industries are taking steps to minimize their negative impact, such as transitioning to renewable energy, improving energy efficiency, and following regulatory guidelines, the crypto and AI industries have largely gone unnoticed.
Despite being a topic of discussion for years, international governments have yet to take any meaningful action. Professor Vered Blass from the Department of Environmental Studies at Tel Aviv University highlights the significant environmental cost associated with the use of AI, including the consumption of raw materials, energy, water, and the emission of various pollutants. This raises concerns about whether tech companies are considering the environmental impact of their actions.
According to Professor Vered Blass from the Department of Environmental Studies at Tel Aviv University, effective regulation can contribute to the improvement of code efficiency, the integration of renewable energy sources, shared responsibility between software and hardware companies within a comprehensive manufacturer’s warranty, achievement of challenging reduction targets, and increased energy efficiency.
The AI industry’s contribution to the climate crisis must be addressed, and companies must be held accountable for the indirect environmental impacts caused by the use of their software services. It is time for the tech industry to lead the way in combating the climate crisis, not contribute to it.
Conlcusion:
The rapid growth of the generative AI industry and the crypto market is having a significant impact on the environment, causing carbon emissions, mining of rare minerals, water, and noise pollution, and contributing to the climate crisis. Despite the efforts of some tech companies to reduce their impact, international governments have done little to address the issue.
The AI industry must be held accountable for the indirect environmental impacts of their software services and work towards reducing their contribution to the climate crisis. Effective regulation and the development of more efficient code, the use of renewable energy, and energy efficiency measures are crucial in addressing the environmental impact of AI. The tech industry must lead the way in combating the climate crisis, not contribute to it.
