The Contrasting Scrutiny Between AI and Crypto: Why Does AI Slip Under the Radar While Crypto Grabs the Spotlight?

In today’s tech-driven world, two big players, Artificial Intelligence (AI) and cryptocurrencies, have taken the stage. They’re both transforming how we live and work, but there’s a bit of a head-scratcher going on: Why is it that when it comes to environmental concerns, we tend to give AI a pass while slamming and bashing cryptocurrencies like Bitcoin for their big energy consumption?

In this article, we will explore the potential reasons behind the differing approaches to AI and cryptocurrencies.

We will explore the findings of a preliminary study conducted by researchers from the University of California (UC), shedding light on the significant water consumption associated with AI, specifically focusing on Microsoft‘s Iowa data center.

Crypto vs. AI

Despite the significant environmental impact, including substantial water and energy consumption associated with AI, investment in the industry continues to surge while the crypto sector faces heightened scrutiny for its high consumption, largely stemming from the widely adopted proof-of-work (PoW) consensus mechanism. 

Nevertheless another reason might be public awareness. For instance, Bitcoin has garnered significant public attention due to its meteoric rise in value and association with cryptocurrencies. As a result, issues related to Bitcoin, including its environmental impact, often make headlines. Moreover, crypto is often a polarizing topic, with strong supporters and critics. The controversy surrounding it, including its environmental impact, may make it a more attractive subject for media coverage.

On the other hand, AI’s environmental impact is less visible and more complex to measure. While AI does require significant computational resources and can contribute to electricity consumption, it’s often applied in data centers, which are designed to be energy-efficient and environmentally conscious. These data centers often use advanced cooling techniques, including liquid cooling, to mitigate their heat output.

Additionally, AI is not as easily relatable to the general public as crypto. In fact, many people use AI-powered services daily without necessarily being aware of the computational resources involved.

Nevertheless, there is a growing awareness of the environmental impact of AI as well. Some media outlets and environmental organizations have raised concerns about the energy consumption of AI models, especially the larger ones like GPT-3 or GPT-4. However, these concerns have not received as much widespread attention as crypto.

The Water Consumption Challenge

The study conducted by UC researchers raises alarm bells regarding the environmental consequences of AI processing, particularly in the case of Microsoft’s Iowa data centers. These data centers play a pivotal role in AI processing and are powered by massive computing resources. For example, Microsoft’s Iowa data center, home to GPT-4, boasts over 285,000 AMD InfiniBand CPU cores and 10,000 GPUs.

The study reveals that from 2021 to 2022, Microsoft’s water consumption in Iowa witnessed a staggering 34% increase, reaching nearly 1.7 billion gallons. What’s striking is that this surge in water usage is closely linked to the proliferation of AI applications. The researchers estimate that a single ChatGPT query can consume approximately 500 milliliters of water, equivalent to a 16-ounce bottle. This water consumption is not solely limited to direct usage but includes indirect sources such as cooling power plants supplying electricity to data centers.

One interesting aspect of the study is the climate advantage of Microsoft’s Iowa data center. Iowa’s cooler climate allows the data center to utilize outside air for cooling most of the year, minimizing water consumption. Water is drawn only when the temperature exceeds a specific threshold. However, during the summer, substantial amounts of water are still required.

Factors Influencing Water Consumption

Several factors contribute to the variability in water consumption for AI processing. The range of 5 to 50 ChatGPT queries per 500 milliliters of water depends on server location and weather conditions. In regions with hotter climates, more water is required for cooling purposes, increasing the environmental impact.

It is worth noting that the study did not find a clear correlation between the complexity of AI queries and water or electricity consumption. This indicates that even routine AI operations can have a substantial environmental footprint.

 Tech Giants’ Responses

In response to the study, Microsoft and OpenAI have emphasized their commitment to environmental sustainability. 

Microsoft highlights its efforts to improve efficiency, invest in renewables, and monitor emissions. OpenAI recognizes the energy and water-intensive characteristics of large AI models and is committed to improving their efficiency.

Google’s Role in the Water Consumption Challenge

The study doesn’t solely point fingers at Microsoft. Google, a tech giant known for its AI initiatives, also experienced a 20% growth in water usage across all its U.S. data centers during the same period (2021-2022). This increase is primarily attributed to the surge in AI processing demands. Google’s water consumption patterns differed across locations, with significant spikes in Las Vegas and Iowa, where the thirst for data center resources was most prominent.

Potential Solutions

Addressing the water-intensive nature of AI systems presents various potential solutions. One strategy involves adopting alternative cooling techniques. Notably, air-cooled systems are gaining popularity due to their heightened energy efficiency and reduced water dependency.

These air-cooled systems employ fans to circulate air across the servers, effectively dissipating the generated heat into the surrounding atmosphere.

Additionally, liquid immersion cooling emerges as another viable option, entailing the submersion of servers in a non-conductive liquid like mineral oil or water to facilitate heat removal.

Another avenue for mitigation lies in the utilization of renewable energy sources to power AI systems. Solar and wind energy can be harnessed to generate electricity, thereby diminishing reliance on fossil fuels and curbing greenhouse gas emissions.

Furthermore, the adoption of water recycling and reuse practices can significantly decrease the overall water consumption of AI systems, ultimately mitigating their impact on local ecosystems and communities.

In addition to the previously mentioned solutions, strategies such as optimizing algorithms for water efficiency, extending hardware lifespans, practicing responsible water management, and promoting policies and regulations that incentivize water footprint reduction can all play crucial roles in addressing the water-intensive nature of AI systems.

These multifaceted approaches not only reduce water consumption but also contribute to the overall sustainability of AI development, aligning it with broader environmental goals.

In summary, the rapid growth of AI comes at a very significant environmental cost, especially regarding water consumption, as highlighted by research like the study from the University of California, focusing on data centers like Microsoft’s in Iowa. However, the stark contrast with the crypto sector, which faces heightened scrutiny due to energy-intensive practices, is still a big question that necessitates an answer.

This discrepancy can be attributed, in part, to how visible and relatable these technologies are to the public. Bitcoin’s headlines and polarized debates have shone a spotlight on its environmental impact, while AI’s complex, behind-the-scenes operation and energy-efficient data center practices have kept it somewhat out of the public eye.

Nevertheless, both AI and crypto industries are gradually recognizing the need for greener alternatives and a balanced approach to technological advancement and environmental responsibility, as it is very crucial to keep moving forward and innovating.