In Drought-Stricken Quilicura, Humans Replace AI for a Day

In one of Chile’s most water-stressed regions, the community of Quilicura has taken on an unprecedented challenge: replacing artificial intelligence with human intelligence for a day to make the hidden water footprint of AI impossible to ignore.

On January 31, the town launched Quili.AI, a human-powered alternative to AI that invited people around the world to submit the everyday questions they would normally ask a machine. Instead of being routed through energy-intensive data centres, each prompt is answered directly by Quilicura residents—drawing on lived experience, cultural knowledge, and human judgment. For 24 hours, no servers, no cloud computing, and no cooling systems will be used.

The response was immediate and far-reaching.

In just one day, more than 25,000 prompts were submitted by participants in 67 countries, resulting in thousands of deeply human exchanges that ranged from practical advice to philosophical reflection. One participant asked for the meaning of life—and was coincidentally matched with a local professor of philosophy. Others sought dating advice, asked about daily life in Quilicura, or simply wanted to confirm, “Are you really a human?”

The initiative was led by Corporación NGEN, an organisation dedicated to restoring, promoting, and educating about the natural heritage of its community, in collaboration with independent global creative partner Tombras, which has a major presence in South America and has been closely following the environmental impact of AI infrastructure as water stress intensifies across the region.

Quilicura is located in the Maipo River Basin, one of Chile’s most water-stressed regions, where communities already face prolonged drought. While AI often feels immaterial to users, its infrastructure has a physical footprint—particularly when it comes to water. In Chile, a single large data center can consume between 1 and 3 million liters of water per day under traditional cooling systems. Globally, AI-related water use is projected to reach billions of cubic meters annually within the next few years.

Corporación NGEN built the platform as a way to turn abstract infrastructure into something human and local. Anyone could ask questions and receive real-time responses from community members — the key difference from AI being that these answers come directly from the people of Quilicura, without consuming a single drop of water.

The experiment highlighted a growing tension: every AI interaction consumes water, indirectly but measurably, through data-centre cooling. While invisible to users, the impact is significant. 

Quili.AI makes this abstract infrastructure tangible. As users submit questions, the platform estimated how much water would have been consumed had the prompt been processed by a conventional AI model—and displayed how much was saved by choosing a human response instead.

The timing is deliberate. In recent years, large-scale viral AI trends have shown how quickly casual prompting can compound. One global wave of AI-generated self-portraits was estimated to consume over 200 million litres of water in under a week—roughly the monthly water usage of a small city.

Organisers emphasised that this initiative was not anti-AI. Rather, it’s a moment of pause—an invitation to prompt responsibly, and consider how these systems should scale in regions already strained by drought. The project aimed to spark dialogue with local officials and policymakers about clearer environmental standards for data centres operating in the Maipo River Basin and beyond.

Global participation was encouraged, and anyone could prompt Quilicura’s residents at Quili.AI on January 31.