Chilean Community Powers Human Chatbot to Expose Hidden AI Water Footprint in Drought-Stricken Region

Human-Powered Chatbot Challenges AI's Environmental Impact

In an initiative to draw attention to the AI water footprint, approximately 50 volunteers from Quilicura, Chile, spent 12 hours operating a fully human-powered chatbot called Quili.AI1

1

. The project, which took place on a Saturday, fielded more than 25,000 requests from users worldwide3

3

. Unlike ChatGPT or Google Gemini, which generate instant responses, this human chatbot deliberately slowed down the process. When one user requested an image of a "sloth playing in the snow," a volunteer responded in Spanish asking them to wait, then delivered a hand-drawn pencil sketch about 10 minutes later1

1

.

Highlighting Water Consumption in Water-Stressed Regions

The initiative spotlights a critical concern: AI data centers consume massive amounts of water for cooling their servers. In Quilicura, some facilities use more than 1 million liters of water every day2

2

. "The goal is to highlight the hidden water footprint behind AI prompting and encourage more responsible use," said Lorena Antiman of environmental group Corporación NGEN1

1

. This municipality at the urban edge of Santiago has become a data center hub, creating tension in a region already grappling with resource consumption challenges. The Maipo River basin, which supplies water to 7 million people including Santiago residents, is one of the most water-stressed regions in the world2

2

.

Environmental Cost of AI Infrastructure Under Scrutiny

The backdrop for this campaign reflects growing debate about the environmental cost of AI infrastructure. Data center computer chips running AI systems require huge amounts of electricity and large volumes of water for data center cooling, with usage varying by location and equipment type1

1

. Cloud computing giants including Amazon, Google, and Microsoft have built or planned multiple facilities in the Santiago region3

3

. Google switched on its Quilicura data center in 2015, claiming it's the "most energy efficient in Latin America" and highlighting investments in wetlands restoration and irrigation projects in the Maipo River basin1

1

. However, the company faced a court challenge over another Santiago-area project due to water usage concerns1

1

.

Encouraging Responsible AI Use Without Rejecting Technology

The volunteers, working in rotating crews on laptops in a community center, answered questions quickly when they drew on local cultural knowledge, such as how to make Chilean sopaipillas1

1

. When stumped, they walked around the room seeking answers from others. "Quili.AI isn't about always having an instant answer. It's about recognizing that not every question needs one," Antiman explained1

1

. She emphasized the project isn't designed to reject AI's "incredibly valuable" uses but to encourage reflection on the environmental costs of casual prompting in drought-affected areas like Chile, which has faced a decade of severe drought that experts link to recent deadly wildfires3

3

. The website informed users how much water they saved by choosing human responses over AI, making the environmental impact tangible and immediate2

2

.