AINews Portal

📖 AI Glossary

22 key AI terms in plain English — each links to the related concepts so you can follow a thread instead of memorizing jargon. JSON is available at /api/glossary.

  • Large Language Model (LLM)

    A neural network trained on huge text corpora to predict and generate language.

  • AI agent

    A model wrapped in a loop that can use tools and take multi-step actions toward a goal.

  • Benchmark

    A standardized test set used to compare model capability.

  • Inference

    Running a trained model to get outputs — what you pay for per token.

  • Token

    The chunk of text (≈¾ of a word) that models read and generate, and that APIs bill by.

  • Attention

    The mechanism that lets a model weigh which parts of the input matter for each output.

  • Fine-tuning

    Further training a model on your own examples to specialize its behavior.

  • Multimodal

    Models that handle more than text — images, audio, video — in and/or out.

  • Embedding

    A list of numbers representing meaning, so similar texts land near each other.

  • Transformer

    The neural architecture behind modern AI, built on the attention mechanism.

  • Diffusion model

    The image/video generation technique that denoises random noise into pictures.

  • MCP (Model Context Protocol)

    An open standard for connecting AI assistants to tools and data sources.

  • RAG (Retrieval-Augmented Generation)

    Fetching relevant documents and feeding them to the model so answers cite real data.

  • Hallucination

    When a model states something false with full confidence.

  • LoRA (Low-Rank Adaptation)

    A cheap fine-tuning method that trains small adapter matrices instead of the whole model.

  • Tool use (function calling)

    Letting a model call external functions/APIs with structured arguments.

  • Context window

    How much text a model can consider at once, measured in tokens.

  • Guardrails

    Checks around a model that block unsafe or invalid inputs and outputs.

  • Open weights

    Models whose trained parameters are downloadable and self-hostable.

  • Reasoning model

    A model trained to "think" in long internal chains before answering.

  • Prompt engineering

    Crafting inputs that reliably get the output you want from a model.

  • Vector database

    A database optimized for finding the nearest embeddings to a query vector.