AI has made life easy; we all know it, right? From answering our questions and creating images to helping us with its suggestions when we lose track of time. But all that smart technology comes with a serious issue: it requires a lot of electricity. Now, researchers at the University of Michigan are working to figure out a way to change that, with $1.83 million in DARPA funding and a goal of cutting AI energy consumption by up to 50%.
Need for So Much Power by AI and How Researchers Plan to Find a Way
AI has become nothing less than salt. Popping up everywhere, from chatbots and image generators to healthcare and scientific research. But what is behind the scenes? Powerful computers working hard to make it all happen, it requires a plenty of electricity to keep going. Data centres used around 420 terawatt-hours of electricity in 2024. And by 2030, it is expected that the figure will be more than double as the demand for AI in every sector blooms. The interesting part is that the solution may not be about building bigger computers but helping the existing ones work smarter. The Researchers at the University of Michigan, the University of Illinois Urbana-Champaign and NAE Systems are on the way with a project called E-CALC to strategise a solution for this.
The project aims to help developers take notes on how much energy AI uses at several stages, from preparing data and training models to running them in everyday applications. The team plans to use physical sensors and software tools to measure energy use more accurately and make its tools freely available to other developers.
Could Smarter AI Save Electricity and Cut Costs?
AI doesn’t just need electricity. Wastage of energy is more than just harmful to the planet. It burns cash on cloud bills, overloads local electric grids and shortens battery life fast. This work is led by University of Michigan professor Mosharaf Chowdhury; the two-year E-CALC project builds on earlier tools that help optimise AI training and compare the energy use of different models. If the project succeeds, then AI could become cheaper to run and more energy efficient without losing what makes it useful.
The idea is simple: to make AI work smarter, waste less power and get more out of every bit of electricity. As AI becomes a bigger part of our lives, this could really become a pretty big win.
For more information, Click Here.
