How can computers understand human language?

The answer

Natural Language Processing (NLP) enables computers to process human language by converting words into high-dimensional numerical vectors (embeddings) where semantic relationships are preserved as mathematical distances. The word 'king' minus 'man' plus 'woman' equals 'queen' in this vector space. Modern transformer-based models like GPT use 'attention mechanisms' to weigh the importance of every word relative to every other word in a passage, allowing them to capture context and nuance across long texts - though they process statistical patterns, not true meaning.