Artificial intelligence could pose a threat to water resources and the environment

Üsküdar Üniversitesi Sağlık Hizmetleri Meslek Yüksekokulu (SHMYO) Sağlık Bilimleri Enstitüsü Assistant Director Dr. Öğr. Üyesi İnci Karakaş, assessed the water resources and environmental issues in artificial intelligence applications.

"Applications have numerous environmental advantages"

Dr. Öğr. Üyesi İnci Karakaş, noting that artificial intelligence applications have numerous environmental advantages, stated, "Through artificial intelligence applications, energy consumption can be analyzed and potential savings areas identified, enabling optimal energy use. Regarding water resources, algorithms that determine water consumption reduce the need for water to a minimum, protecting limited usable water sources, and smart irrigation systems can develop irrigation systems customized for specific crops." she said. [caption id="attachment_160214" align="aligncenter"] Dr. Öğr. Üyesi İnci Karakaş[/caption]  

"Models such as GPT-3 and BLOOM can lead to significant water and energy consumption"

On the other hand, examining recent research on artificial intelligence, Dr. Karakaş noted that various models such as GPT-3 and BLOOM can lead to significant water and energy consumption, and continued: "A study conducted at the University of California estimates that with increasing artificial intelligence applications, 4.2-6.6 billion cubic meters of water could be consumed by 2027. Due to the rapidly growing population and depleting natural water sources, difficulty in accessing fresh and usable water is one of humanity's most urgent shared challenges, which makes this situation quite concerning. It would be important from an environmental perspective to clearly specify the amount of water used in cooling systems to prevent equipment overheating during the development and operation of various artificial intelligence models. In addition to the water footprint indicating the amount of fresh water used to produce a product or service, the carbon footprint must also be presented to create environmentally sustainable artificial intelligence models."

"Drinkable and usable water resources are limited"

Noting that drinkable and usable water resources are limited and unevenly distributed globally, Dr. Öğr. Üyesi İnci Karakaş stated, "Water drawn from underground or surface water sources for agricultural or urban purposes, as well as for artificial intelligence applications, can reduce available water resources. After water is extracted from underground and surface water sources, changes in water quality contribute to water stress levels for subsequent uses. Global water resources are proving inadequate both in quantity and quality, facing the risk of water scarcity." she said.

"Water scarcity affects 4 billion people"

Dr. Öğr. Üyesi İnci Karakaş also stated that severe water scarcity currently affects 4 billion people, approximately two-thirds of the global population, for at least one month each year, and added, "Unless integrated and inclusive approaches are brought to the global water crisis, research emphasizes that nearly half of the world's population could face serious water shortages by 2030 and approximately one in four children globally will be living in areas facing severe water scarcity by 2040. Additionally, the location and time where artificial intelligence applications are trained and operated may be important in terms of water consumption." she informed.

"Insufficient awareness has been created"

Also noting that insufficient awareness has been created regarding water resources used by artificial intelligence applications, Dr. Öğr. Üyesi İnci Karakaş stated, "Since the amounts of water consumed during the development and operation of applications are not clearly specified, ensuring water sustainability may be very difficult. The failure to specify the water footprint may also create a potential barrier to environmentally sustainable artificial intelligence applications in the future. Evaluating water footprint and carbon footprint together is important for sustainable artificial intelligence applications." she said. Mentioning that artificial intelligence applications can be used in environmental science, health, education and research, scientific publishing, digital assistants, transportation and logistics, finance and translation, Dr. Öğr. Üyesi İnci Karakaş concluded her remarks as follows: "The environmental risks that artificial intelligence applications, used in many areas of our lives, could create must be revealed. Since artificial intelligence programs are complex, they require more energy than other computing systems. However, determining exactly how much energy artificial intelligence tools consume is extremely difficult.''

"The effects of global warming could be felt more severely"

In a study conducted in 2022, the BLOOM model's training resulted in estimated carbon dioxide emissions of 24.7 tons, and equipment production and model operation resulted in estimated carbon dioxide emissions of 50.5 tons. Carbon dioxide is seen to account for the largest share among greenhouse gas emissions. With increasing industrialization and population growth, artificial intelligence applications also raising carbon dioxide emissions could increase atmospheric temperature, making the effects of global warming felt more severely. Due to global warming, sea level rise, glacial melting, severe winds and heavy rainfall formation, combined with air quality deterioration, declining natural water sources, water scarcity and increased epidemic diseases could occur."