The Dictionary of AI Coding is a comprehensive reference resource created by Matt Pocock that translates artificial intelligence and machine learning terminology into plain English for developers. The repository is built in TypeScript and serves as an educational glossary designed to demystify AI coding concepts that often feel gatekept by industry jargon. The project addresses a real problem: the vocabulary surrounding AI development can feel deliberately obscured, creating barriers to entry for developers who want to understand how these systems actually work.
The dictionary is organized into seven major sections covering different aspects of AI development. Section 1 covers foundational model concepts including AI itself, models, parameters, training, inference, tokens, and token prediction mechanics. Section 2 addresses sessions, context windows, and conversation turns, explaining how stateless and stateful interactions work. Section 3 focuses on tools and environments, including filesystems, tool calls, and sandboxing. Section 4 examines failure modes like hallucination, sycophancy, and attention degradation. Section 5 covers handoffs between systems and agents. Section 6 discusses memory systems and steering mechanisms. Section 7 explores patterns of work including human-in-the-loop processes, automated checks, and prototyping approaches.
The README excerpt reveals the project's philosophy: rather than assuming readers already understand AI concepts, it explains each term with concrete examples and usage scenarios. For instance, the entry on "AI" itself traces how the term has shifted meaning across decades, from symbolic reasoning in the 1950s through expert systems to modern large language models. This historical context helps readers understand why conversations about AI often talk past each other. Similarly, the "Model" entry clarifies that models are stateless parameter sets doing next-token prediction, and that agentic behavior comes from the harness orchestrating multiple predictions, not from the model itself.
According to GitGenius activity tracking, the repository has seen consistent engagement with a median issue and pull request response latency of 0.0 hours and a mean of 100.9 hours across four tracked items. Matt Pocock is the primary maintainer with 9 recorded events, while contributors Haltarys and ubidev have each contributed 1 event. The most active issue label is needs-triage with 3 occurrences. The repository is classified across multiple domains including AI terminology, coding glossary, machine learning, artificial intelligence, developer reference, technical dictionary, programming concepts, language models, software engineering, and educational resources.
The project connects to other repositories in Matt Pocock's ecosystem through overlapping contributors, linking to course-video-manager, sandcastle, and evalite. The README indicates the dictionary is generated from source files, with a note that it is a generated file that should not be edited directly. The project also promotes an associated newsletter at aihero.dev with over 62,000 subscribers, positioning the dictionary as part of a broader educational initiative around AI engineering skills and practices.
