# What Is ChatGPT Doing … and Why Does It Work?  ## Metadata - Author: [[stephenwolfram.com]] - Full Title: What Is ChatGPT Doing … and Why Does It Work? - Category: #articles - Summary: insert summary - My notes: - Document Tags: [[llms]] - Summary: The article discusses how neural nets work, particularly in the context of ChatGPT, a language model designed to mimic human language. The neural net corresponds to a mathematical function with billions of terms, and it "recognizes things" by generalizing from examples it has been shown. ChatGPT uses a transformer architecture to learn the nested-tree-like syntactic structure of human language. The article also discusses the challenges of training neural nets and the need for data augmentation. Ultimately, the neural net in ChatGPT has implicitly discovered the regularities in language that make it possible to generate human-like text. - URL: https://writings.stephenwolfram.com/2023/02/what-is-chatgpt-doing-and-why-does-it-work/ ## LLM Chats ## NotebookLM ## LLM Audio ## Highlights > Are our brains using similar features? Mostly we don’t know. But it’s notable that the first few layers of a neural net like the one we’re showing here seem to pick out aspects of images (like edges of objects) that seem to be similar to ones we know are picked out by the first level of visual processing in brains. ([View Highlight](https://read.readwise.io/read/01k8ne9a9agqd8ek8rhaz66xkt))