The mattpocock/dictionary-of-ai-coding repository is a comprehensive resource designed to demystify the jargon and terminology commonly encountered in AI coding and engineering. Its primary purpose is to provide clear, plain-English explanations for the specialized vocabulary used in the field of artificial intelligence, particularly as it relates to coding and working with large language models (LLMs). The repository aims to make AI concepts accessible to a broader audience, including developers who may be new to the domain or those seeking to deepen their understanding without getting lost in technical complexity.
The repository is written in TypeScript and is structured around a dictionary format, with terms organized into thematic sections. Each section covers a different aspect of AI coding, such as models, sessions, context windows, tools and environments, failure modes, handoffs, memory and steering, and patterns of work. For example, the section on "The Model" explains foundational concepts like AI, model, parameters, training, inference, token, and next-token prediction. These explanations are crafted to clarify not only what each term means but also how it fits into the broader context of AI development and usage.
One of the key features of the repository is its focus on practical understanding. It addresses common sources of confusion in AI coding, such as why context degrades, why billing can be unexpectedly high, and why prompts may behave inconsistently. By translating technical jargon into everyday language, the dictionary helps users navigate the complexities of AI systems, understand their limitations, and make informed decisions about their use. The repository also highlights the importance of precise terminology, encouraging users to replace vague references to "AI" with more specific terms like "model," "harness," or "agent" to improve communication and troubleshooting.
The dictionary is not static; it is generated from a collection of markdown files and internal curriculum documents, ensuring that the content remains up-to-date and relevant. The README provides a detailed table of contents, allowing users to quickly locate terms and concepts of interest. Each entry includes not only definitions but also usage examples and guidance on avoiding common misunderstandings. For instance, the explanation of "AI" discusses how the term has evolved over time and why it is often misused in technical discussions, advocating for greater specificity.
In addition to the dictionary itself, the repository connects users to further resources, such as the AI Hero newsletter, which offers ongoing updates and insights for developers interested in staying ahead in the rapidly evolving field of AI engineering. The repository is part of a broader effort to foster transparency and accessibility in AI, countering the tendency for jargon and complexity to obscure understanding and limit participation.
Overall, mattpocock/dictionary-of-ai-coding serves as a valuable educational tool for anyone working with or interested in AI coding. Its clear explanations, practical focus, and structured organization make it an ideal starting point for learning the vocabulary of AI, understanding the mechanics behind key concepts, and navigating the challenges of modern AI development. By breaking down barriers to understanding, the repository empowers developers to engage more confidently and effectively with AI technologies.
