Artificial intelligence is changing what software can do, but every model also depends on physical infrastructure. Training and serving models require electricity, cooling, water, hardware, and mineral supply chains. As demand grows, the environmental cost of AI becomes a question for everyone who builds, buys, governs, or uses these systems.

This session looks beyond impressive demos to ask better questions: Which applications create meaningful public value? How should we account for energy and water use? What can efficiency, smaller models, renewable power, and transparent reporting change? The goal is not to reject the technology or accept its impacts as inevitable, but to make climate responsibility part of the design brief.