Latency hiding is a technique used in computing to mask the delay between a request and its completion, improving the perceived performance of a system. It is achieved through methods such as prefetching, caching, and parallel execution to ensure that the system remains responsive and efficient despite inherent delays.