By Michael Scott
Artificial intelligence joins Bitcoin mining as another economic activity causing environmental scourge warranting condemnation. The challenge is associated with the high energy demands to run the computing equipment.
The intense energy consumption translates to higher water volume to cool the servers answering the prompts fed. The emergence of superior AI and multiple firms eyeing the sector indicates the thirst would only grow bigger everyday.
AI systems often feature intense computational activity with models such as the OpenAI’s ChatGPT demanding colossal processing power. Similarly, the AI models require substantial electrical power for the servers. The serves are located within data centers sprouting daily, placing a huge demand for water.
Widespread Data Centers Consuming Water and Energy Amounts
The awareness of the adverse environmental effects of AI systems explains the move by Microsoft and Google to lead other tech players to implement corrective measures to reduce the environmental impact burden. The initiative collaborates with the broader tech industry’s move to minimize resource utilization. Also, American multinationals support the experts’ call to raise awareness of the adverse impact of uncontrolled expansion in the tech industry.
A Tuesday, September 5 publication by the Associated Press (AP) estimated that AI systems of ChatGPT magnitude drain 500 milliliters of water whenever a user poses 5 to 50 prompts. The AP publication posed that with the Microsoft-backed OpenAI realizing the fastest growth since internet creation, the drain on water supplies is enormous. The water requirements prompt experts and environmentalists to raise the alarm concerning AI systems threatening to drain the water supply.
Powering AI Supercomputers Draining Water Supplies
The publication reveals that ChatGPT relied upon the data centers in western Des Moines, Iowa. The location enabled the center to leverage water supplied from the Raccoon and Des Moines rivers. The AP publication cited the admission by Microsoft president Brad Smith that though California-based engineers developed GPT-4, it was created in Lowa.
The unveiling of ChatGPT triggered an accelerated AI revolution that brings with it intense water consumption in running the power-hungry hardware. The equipment needs water to generate energy powering the hardware and huge units to cool the hardware.
Microsoft’s official report concerning water consumption shows a 34% spike in 2022, primarily fuelled by the increased AI-related computing demands.
Google disclosed a 20% 8ncrease in water consumption in 2022. Translating the percentage into real data shows that the two companies accounted for draining an additional 1.7 billion gallons.
OpenAI appears to identify ideal locations for its data centers selectively. Nonetheless, residents would soon feel the impact, unaware of the contributing factor. A review of residential rates within Des Moines Water Works reveals 80% growth since 2017. Axios reported in April last year that 1000 gallons of water cost $5.19.
Accelerated AI Technology Imposing Unintended Consequence to Environment
Experts point out that unintended consequences arise from accelerated AI technology advancements, which demand more computing power.
Environmentalist Shaolei Ren informed AP that most individuals are unaware of resource usage in running ChatGPT. The researcher laments that being unaware of resource usage hinders their capability to assist in conserving the resources.
In their recent reports, tech companies have disclosed comprehensive plans to lower environmental impacts.
Google indicated that the data center’s energy efficiency is 60% above the industry average. Microsoft unveiled an initiative encouraging transparency and efficiency improvements within all leased locations. The Smith-led firm targets 100% carbon-free energy in 7 years.
In last year’s sustainability report, Microsoft restated its focus on housekeeping duties to realize the 2030 carbonless energy goals. The objective targets transforming Google into a water-positive yet carbon negative and zero wastage by 2030.
Google Commitment to Net-zero Emissions
Google’s 2023 environmental report mirrors the stance adopted by Microsoft. In particular, Google targets net-zero emissions by 2030. The multinational mandates rapid reduction measures across its value chain and operations. Doing so implies eliminating 10.2 million tons of carbon emitted in 2022.
Google observes that efficiency practices can slow the increase in energy required to power the accelerated demand for AI. Nonetheless, efficiency practices can hardly reverse the environmental burden imposed by the power-hungry and water-thirsty AI.
The awareness of the drain on water supplies necessitates diligent pursuit of renewable energy blended with alternative cooling techniques. Also, growth in AI capabilities should feature transparency in the disclosure of resource usage. Careful resource usage management is paramount to realizing a sustainable quench of AI’s thirst.