A new study from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) demonstrates a novel approach to training AI models that significantly reduces energy consumption. The method, called ‘Lazy Training,’ selectively activates only the necessary parts of a neural network for a given task, rather than running the entire model. Initial results show a potential reduction in computational energy use by up to 80% for certain inference tasks, without a corresponding loss in accuracy. Researchers highlight that this technique could make deploying large AI models more sustainable and cost-effective, particularly for mobile and edge computing devices where power is limited. The full implications for training efficiency are still under investigation. Read the full article at: https://technologyreview.com/2024/05/15/ai-energy-efficiency-lazy-training/