Catastrophic Forgetting

This is a problem in machine learning where a model, when learning new tasks, suddenly forgets information it previously learned from older tasks. It essentially 'overwrites' old knowledge with new, making it difficult for the model to retain a diverse set of learned skills over time.