Why Large Language Models Are Just Mathematical Functions: A Rational Perspective
The article argues that large language models are fundamentally mathematical functions that model human language, emphasizing their role as simplified representations, explaining their structural nature, sources of errors, the importance of prompts as boundary conditions, and the need for clear usage assumptions to avoid anthropomorphic misconceptions.