The only AI glossary you’ll need this year
Original reporting by TechCrunch

The language of artificial intelligence refers to the rapidly expanding lexicon of technical terms, acronyms, and specialized concepts defining the cutting edge of modern technology. As AI rewrites industries and transforms daily life, it simultaneously invents a whole new vocabulary to describe its methods and capabilities. From Large Language Models (LLMs) and Reinforcement Learning from Human Feedback (RLHF) to the nuances of diffusion and distillation, this specialized jargon can leave even seasoned tech professionals and investors feeling uncertain, creating a significant barrier to understanding the ongoing AI revolution. The rapid pace of innovation seems matched only by the speed at which new terms emerge, making it challenging for anyone to stay truly informed.
Navigating the Jargon
This comprehensive and authoritative glossary serves as an essential guide, meticulously crafted to demystify these critical terms. We provide clear, plain-English definitions for the most prevalent AI concepts, equipping readers whether they are actively developing new systems, strategically investing in the sector, or simply striving to remain informed by following industry news and analysis. Far from a static compilation, this resource is a living document, meticulously updated to reflect the dynamic evolution of the field. It ensures that you always have access to current and understandable explanations of the powerful technologies shaping our collective future, empowering you to speak the language of AI with confidence.
This living glossary serves as a vital compass in the rapidly expanding universe of artificial intelligence. As we've detailed, the nuanced language of AI — from LLMs and RAG to AGI and recursive self-improvement — is not just jargon; it’s the blueprint of a transformative era. Understanding these concepts empowers a clearer grasp of current advancements and the foundational challenges developers face daily, whether grappling with hallucinations or optimizing for inference through memory caching.