New materials and nanomagnets might accelerate the benefits of artificial intelligence.
Researchers from Imperial College London have shown for the first time that networks of nanomagnets can do AI-like processing. The approach might drastically cut the worldwide doubling of artificial intelligence’s energy use every 3.5 months. It might be used for ‘time-series prediction’ jobs, such as forecasting and controlling insulin levels in diabetic patients.
Nanomagnets are super-efficient computers that process information in the form of a “magnon” wave. Each magnet influences the state of its neighbors: This implies that far less energy has wasted and that information processing and storage may be performed concurrently, unlike in traditional computers. They might be included in conventional computers to enhance the energy efficiency of intensive processing operations, such as AI training.