TechCrunch has published a glossary explaining commonly used artificial intelligence terminology, aiming to help readers keep pace with the rapidly evolving vocabulary surrounding AI development. The piece is framed as a "living document" that will be updated as the field changes, reflecting how quickly new jargon—such as "opaque recurrence," a reasoning technique recently introduced in OpenAI's Astra model—enters mainstream tech discussion and reportedly unsettled AI safety researchers.

The glossary defines terms including AGI (artificial general intelligence), which OpenAI and Google DeepMind describe in slightly different ways as AI matching or exceeding human capability across most tasks; AI agents, autonomous tools that perform multistep tasks such as booking travel or writing code; API endpoints, the interfaces allowing software and AI agents to interact with other programs; and chain-of-thought reasoning, where AI models break complex problems into intermediate steps to improve accuracy, similar to how a person might work through a maths problem on paper.