A new study from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) demonstrates a significant advancement in making AI systems more energy-efficient. The research focuses on developing specialized hardware, specifically a new type of photonic chip, that can perform complex AI computations using light instead of electricity. This approach, known as optical neural networking, promises to drastically reduce the power consumption and heat generation associated with large-scale AI models and data centers. The experimental chip successfully ran a machine learning model for image recognition tasks, showing comparable accuracy to traditional electronic systems while using a fraction of the energy. Researchers highlight that this technology could enable more powerful and sustainable AI applications in the future, from autonomous vehicles to large language models, by overcoming current thermal and power limitations. For the full details, read the complete article at the source.