How you write an AI prompt dramatically changes its carbon footprint | Yankai Jiang | TEDxBoston

Data centers are "hungry" and "thirsty", now how do we keep them from getting hangry? Just as we might talk about balance and optimization for human performance in the workplace, the growing fleet of digital colleagues can work smarter, not harder. How we design and operate these tools greatly affects the efficiency. One video generation request produces five times as much carbon as a single text query, and adding a sentence to an AI prompt can reduce its resource demand by up to half. A data center in Arizona running on coal power is a fundamentally different environmental proposition than one in Finland cooling with surrounding seawater and batching jobs when renewable energy peaks. Sustainable computing approaches extend the value and capacity of our digital infrastructure. Yankai (Luke) Jiang is a PhD researcher in computer science focused on sustainable data center design, with particular expertise in the environmental tradeoffs between carbon emissions and water consumption in AI computing infrastructure. His research developed Water-Wise, a system that jointly optimizes workload scheduling both for carbon and water impact, and Eco-Life, a framework for intelligently combining multiple generations of computing hardware to reduce sustainability costs. This talk was given at a TEDx event using the TED conference format but independently organized by a local community. Learn more at https://www.ted.com/tedx