Introduction
Artificial intelligence is reshaping the world—and inventing a new vocabulary to describe how it’s doing so. Whether you’re building with AI, investing in it, or simply trying to keep up with the latest TechCrunch articles or podcasts, terms like LLMs, RAG, RLHF, and others can leave even seasoned tech professionals feeling uncertain. This glossary provides plain-English definitions for the AI terms you’re most likely to encounter. As the field evolves rapidly—with 2026 bringing deeper integration of AI into daily life and enterprise—we update this living document regularly to reflect the latest developments.
AGI (Artificial General Intelligence)
Artificial general intelligence (AGI) remains a nebulous concept, but generally refers to AI that matches or exceeds human capabilities across many—if not most—tasks. OpenAI CEO Sam Altman once described AGI as the "equivalent of a median human that you could hire as a co-worker." In contrast, OpenAI’s charter defines it as "highly autonomous systems that outperform humans at most economically valuable work." Google DeepMind’s characterization is slightly different, viewing AGI as "AI that’s at least as capable as humans at most cognitive tasks." Confused? You’re not alone—even leading AI researchers struggle to agree on a precise definition. In 2026, the debate continues as models approach human-level performance in more domains, though true AGI remains elusive.
AI Agent
An AI agent is a tool that employs AI to perform a series of tasks on your behalf—going beyond what a basic AI chatbot can do. Examples include filing expenses, booking tickets or restaurant reservations, and even writing or maintaining code. However, as previously discussed, this emergent space has many moving parts, so “AI agent” may mean different things to different people. By 2026, the infrastructure for multi-step, autonomous workflows has matured significantly, with agents now commonly orchestrating complex tasks like managing supply chains or coordinating business travel.
API Endpoints
Think of API endpoints as “buttons” on the back of a software system that other programs can press to trigger actions. Developers use these interfaces to build integrations—for example, allowing one application to pull data from another or enabling an AI agent to control third-party services without manual input. Most smart home devices and connected platforms offer these hidden buttons, even if ordinary users never see them. As AI agents have grown more sophisticated through 2025 and into 2026, they increasingly discover and use these endpoints autonomously, unlocking powerful (and sometimes unexpected) automation possibilities—from syncing calendars to adjusting building HVAC systems.
Chain of Thought
Given a simple question like “Which animal is taller, a giraffe or a cat?” the human brain answers almost instantly. But many problems require writing down intermediate steps—for example, if a farmer has 40 heads and 120 legs from chickens and cows, you’d need a simple equation to find the answer (20 chickens and 20 cows). In AI, chain-of-thought reasoning means breaking a problem into smaller, sequential steps to improve output quality. Large language models use this technique to tackle complex tasks such as math word problems, logical deduction, and coding. While it typically takes more time, by 2026, specialized models and inference optimizations have made chain-of-thought faster and more reliable for real-time applications like customer support and legal document analysis.
via TechCrunch AI