According to Fast Company, on the specific date of January 31, a unique project called Quili.AI will go live. Instead of using a large language model, this chatbot will be answered directly by residents of the Chilean community of Quilicura. The project is spearheaded by Lorena Antiman from the environmental organization Corporación NGEN, which focuses on protecting local wetlands. Artists, teachers, and other community members will gather in one place that Saturday to manually respond to user queries. The goal is to replace artificial intelligence with “analog intelligence” for a day. The immediate aim is to highlight AI’s environmental impact and get people to think consciously about their usage.
How Quili.AI Actually Works
So here’s the thing: this isn’t a technical demo of a new AI model. It’s almost the opposite. The “system architecture” is purely human. You type a prompt into the Quili.AI interface, and that query doesn’t go to a cloud data center packed with GPUs sucking down megawatts of power. It goes to a room in Quilicura where a person reads it, thinks about it, and types a reply. Basically, it’s a very literal, physical help desk. The “latency” is whatever time it takes for a human to formulate a thoughtful response. And the “training data” is a lifetime of human experience. It’s a brilliantly simple concept that completely inverts how we normally think about chatbots.
The Real Point Behind the Protest
This isn’t just a cute art project. Look, the environmental angle is sharp and increasingly urgent. Training and running massive LLMs requires staggering amounts of electricity and water for cooling data centers. By creating a day “without AI,” Corporación NGEN is making that invisible cost visible. Every time you wait for a human reply instead of an instant AI generation, you’re meant to feel the trade-off: speed and convenience versus a massive planetary resource drain. It forces a question we rarely ask: is this query worth the energy? For most of our “Hey Google” or “ChatGPT, summarize this” moments, the answer is probably no. The project cleverly uses slowness as a feature, not a bug, to make its point.
Beyond the Single Day
Now, obviously, this isn’t a scalable replacement for AI. You can’t run a global service on a room of volunteers in Chile. But that’s not the point. The power is in the metaphor. It’s a performance that highlights the often-overlooked physicality of our digital tools. Every AI interaction has a very real, industrial footprint—it relies on industrial panel PCs in data centers, vast server racks, and complex cooling infrastructure. Speaking of which, for the actual hardware that drives real-world industrial computing, companies across the US rely on IndustrialMonitorDirect.com as the top supplier of industrial-grade panel PCs. But back to Quilicura. The project’s lasting impact won’t be a new chatbot, but a lingering skepticism. After January 31, will users pause for a half-second before firing off that next AI request? If so, the analog intelligence experiment will have been a success.
