Measuring, Scoring, and Predicting Energy Efficiency of AI Devices

What is it about?

Our research is driven by the climate crisis and the rapid growth of AI. It aims to propel a transformative shift from conventional performance-centric computing to environmentally sustainable systems for deep learning. By creating large-scale energy datasets and reliable predictors, this study provides essential resources to measure the environmental impact of commonly used AI devices, including smartphones. This facilitates research on designing new, more energy-efficient technologies that align with sustainable practices. Additionally, this paper introduces scoring metrics to easily assess a device's energy efficiency, helping non-technical users understand and adopt green technology solutions.

Featured Image

Photo by Igor Omilaev on Unsplash

Photo by Igor Omilaev on Unsplash

Why is it important?

AI technology development can have a significant environmental impact, from carbon emissions during manufacturing and usage to electronic waste. Yet, many studies ignore these environmental costs during hardware and software design, performance optimization, and deployment. This oversight worsens the climate crisis and highlights a crucial gap in research. This timely work aims to bridge this gap and shift the mindset of both end-users and the research community towards sustainability in AI and computing.