The increasing popularity of AI tools such as OpenAI's ChatGPT has raised concerns about their environmental impact. A recent study has revealed that AI models have a substantial water footprint due to the significant amounts of water needed to sustain data centres and train these models.
The water footprint of AI refers to the amount of water used in the process of generating electricity and providing cooling for data centres that run AI models. There are two components to the water footprint: direct water consumption and indirect water consumption.?
Direct water consumption involves the water evaporated or discharged as waste during the cooling of data centre servers. Indirect water consumption refers to the water used to produce electricity that powers the data centre servers.?
The water footprint of AI can vary based on factors like the size and type of AI model, the efficiency and location of the data centre, and the source of electricity generation.
A recent study titled "Making AI Less 'Thirsty:' Uncovering and Addressing the Secret Water Footprint of AI Models" found that training large AI models like GPT-3 can consume up to 700,000 litres of clean freshwater.?
According to The Down to Earth report, this amount of water is equivalent to producing 370 BMW cars or 320 Tesla electric vehicles. Additionally, using AI chatbots like ChatGPT in conversations can consume up to 500 ml of water for 20-50 questions and answers.?
Considering that ChatGPT has over 100 million active users engaging in multiple conversations, this water usage adds up. The upcoming GPT-4, which is expected to be even larger, is predicted to further increase water consumption, although specific estimates are challenging due to limited data availability.?
While AI activities occur digitally, the physical storage and processing of data in data centres generate heat, requiring water-intensive cooling systems. These systems use pure freshwater and also require significant water for power generation.
The water footprint of AI contributes to water scarcity by consuming significant amounts of freshwater for cooling AI infrastructure, which strains limited water resources. This exacerbates the global issue of water scarcity.
The extraction of freshwater for AI operations can harm aquatic biodiversity, negatively impacting the environment. Additionally, the energy required for water treatment and transport for AI operations contributes to carbon emissions and climate change, further exacerbating environmental concerns.
Moreover, diverting water for AI operations can lead to unsustainable resource management. This diversion of water may hinder access to water for human consumption, agriculture, and other critical needs, potentially exacerbating water scarcity issues for communities.
Furthermore, the water-intensive nature of AI can have equity and social implications. Vulnerable communities that rely on limited water supplies for their livelihoods are disproportionately affected by water scarcity. The water requirements of AI could worsen existing inequities by diverting water away from communities that need it the most.
It is also important to consider long-term sustainability. The expanding AI industry could place additional strain on water resources if the water footprint issue is not addressed. Therefore, addressing the water footprint is crucial for ensuring the long-term sustainability of both AI development and water availability.
Utilizing renewable energy sources such as wind or solar power reduces the water required for generating electricity.
Implementing water-efficient cooling systems like air cooling or direct-to-chip liquid cooling in data centres can minimize water usage. Further, developing water-efficient algorithms by optimizing processes and reducing computational power can help conserve water.?
Extending the lifespan of hardware and promoting responsible water management practices, such as recycling wastewater and implementing water-efficient landscaping, are additional measures.?
And adopting policies and regulations that incentivize or mandate water footprint reduction through standards, targets, or taxes can further drive the sustainable use of water in AI.