Category: Encyclopedia

Open Source AI — AI models released with open weights that anyone can download, modify, and deploy. Leaders: Meta (Llama), Mistral, DeepSeek. Benefits: privacy, customization, no vendor lock-in. Run locally with Ollama or LM Studio. Why It Matters Understanding Open Source AI is essential for anyone building or evaluating AI systems. As AI tools proliferate,…

LLM (Large Language Model) — A neural network trained on massive text datasets that can generate, understand, and reason about language. Examples: GPT-4, Claude, Gemini, Llama. Sizes range from 7B to 1.8T parameters. The foundation of modern AI. Why It Matters Understanding LLM is essential for anyone building or evaluating AI systems. As AI tools…

Token — The basic unit AI models process. Roughly 1 token = 0.75 words in English. ‘Hello world’ = 2 tokens. Pricing, context windows, and speed are all measured in tokens. GPT-4o: $2.50/1M input tokens. Claude Sonnet: $3/1M input tokens. Why It Matters Understanding Token is essential for anyone building or evaluating AI systems. As…

Hallucination — When an AI model generates confident but factually incorrect information. It sounds right but isn’t. Causes: training data gaps, pattern matching without understanding. Mitigation: RAG, grounding, fact-checking layers, and trust scoring. Why It Matters Understanding Hallucination is essential for anyone building or evaluating AI systems. As AI tools proliferate, knowing the fundamentals helps…

Agentic AI — AI systems that can plan, use tools, and take actions autonomously. Unlike chatbots that just respond, AI agents can browse the web, write code, query databases, and chain multiple steps together. MCP is the protocol that connects agents to tools. Why It Matters Understanding Agentic AI is essential for anyone building or…

Inference — Running a trained AI model to generate predictions or responses. Training creates the model; inference uses it. Speed matters — Groq claims 500+ tokens/second. Cost matters too — inference pricing determines your AI bill. Why It Matters Understanding Inference is essential for anyone building or evaluating AI systems. As AI tools proliferate, knowing…

Context Window — The maximum amount of text an AI model can process at once, measured in tokens. GPT-4o: 128K tokens. Claude: 200K tokens. Gemini: 1M tokens. Larger context windows let models work with longer documents but cost more and can be slower. Why It Matters Understanding Context Window is essential for anyone building or…

RLHF (Reinforcement Learning from Human Feedback) — A training technique where human evaluators rank AI outputs, and the model learns to produce responses humans prefer. This is how ChatGPT became conversational. The human feedback loop makes models more helpful, harmless, and honest. Why It Matters Understanding RLHF is essential for anyone building or evaluating AI…

Prompt Engineering — The practice of crafting inputs to AI models to get better outputs. Techniques include few-shot examples, chain-of-thought reasoning, system prompts, and structured formatting. The difference between a useless AI response and a perfect one. Why It Matters Understanding Prompt Engineering is essential for anyone building or evaluating AI systems. As AI tools…

Embeddings — Embeddings convert text, images, or data into numerical vectors that capture meaning. Similar concepts get similar vectors. This enables semantic search, recommendation systems, and RAG pipelines. Key providers: OpenAI, Cohere, Voyage AI. Why It Matters Understanding Embeddings is essential for anyone building or evaluating AI systems. As AI tools proliferate, knowing the fundamentals…

MCP (Model Context Protocol) — MCP is an open protocol that standardizes how AI models connect to external tools and data sources. Think of it as USB for AI — any model can plug into any tool through a standard interface. Created by Anthropic, now used across the AI ecosystem with 15,000+ servers. Why It…

Fine-Tuning — Fine-tuning adapts a pre-trained AI model to a specific task by training it on additional domain-specific data. Instead of building from scratch, you take GPT-4 or Llama and specialize it for your use case — medical records, legal documents, customer support. Why It Matters Understanding Fine-Tuning is essential for anyone building or evaluating…

RAG (Retrieval Augmented Generation) — RAG combines AI language models with external knowledge retrieval. Instead of relying only on training data, RAG systems search a database or documents first, then use that context to generate more accurate, current answers. Essential for enterprise AI where accuracy matters. Why It Matters Understanding RAG is essential for anyone…

Understanding the Model Context Protocol The Model Context Protocol (MCP) is an open standard that defines how AI agents discover, authenticate with, and invoke external tools and data sources. Developed to solve the fragmentation problem in AI tooling, MCP provides a universal interface between large language models and the services they need to accomplish real-world…