Catastrophic forgetting is a term used in artificial intelligence and also applies to areas such as automation and Industry 4.0. It describes a problem that occurs when an artificial intelligence (AI) suddenly loses or "forgets" what it has previously learnt when learning new tasks.
Imagine a robot first learns how to tighten screws. Later, it is taught how to hammer nails. If the robot completely forgets how to tighten screws when learning the new skill, this is known as catastrophic forgetting. This makes the AI less useful, as it cannot retain the knowledge it previously had.
This problem is particularly critical when AI systems are required to flexibly perform many different tasks - for example in a smart factory where machines are constantly being used for different tasks.
Catastrophic forgetting is therefore a key issue in the development of modern AI solutions. Innovative methods are being developed to ensure that AI systems not only learn new things, but also retain their old knowledge - similar to humans, who do not forget everything they have learnt before when learning new skills.
Deutsch 
English (UK)