AI Glossary for Beginners: 50 Terms You Need to Know

The world of AI comes with its own vocabulary, and the jargon can make a simple idea sound impenetrable. This glossary defines 50 of the terms you will run into most often, in plain English, grouped so related ideas sit together. It is built for newcomers — founders, marketers, and curious readers — not researchers, so definitions favor clarity over precision. Read it start to finish or use it as a reference when a term trips you up. Where two terms are easy to confuse, we put them side by side on purpose.

The basics

• Artificial intelligence (AI): Software that performs tasks we associate with human intelligence, like understanding language or recognizing images.
• Machine learning (ML): A way of building AI where a computer learns patterns from data instead of being programmed with explicit rules.
• Deep learning: A type of machine learning that uses neural networks with many layers to learn complex patterns.
• Neural network: A computing system loosely inspired by the brain, made of connected units that pass signals and adjust as they learn.
• Algorithm: A defined set of steps a computer follows to solve a problem or process data.
• Model: The trained system that takes an input and produces an output, such as text or a prediction.
• Dataset: The collection of examples used to train or test a model.
• Data: The raw information — text, images, numbers — that models learn from and operate on.

How models learn

• Training: The process of teaching a model by showing it data and adjusting its internal settings to reduce errors.
• Supervised learning: Training on labeled examples, where each input comes with the correct answer.
• Unsupervised learning: Training on unlabeled data, where the model finds patterns or structure on its own.
• Reinforcement learning: Training where a model learns by trial and error, guided by rewards for good outcomes.
• Parameter: An internal value a model adjusts during training; large models have billions of them.
• Weight: A specific kind of parameter that controls how strongly one signal influences another.
• Backpropagation: The standard method for adjusting a neural network's weights based on its mistakes.
• Overfitting: When a model memorizes its training data too closely and performs poorly on new inputs.
• Generalization: A model's ability to perform well on data it has not seen before — the opposite goal of overfitting.
• Fine-tuning: Further training of an existing model on a narrower dataset to specialize it for a task.

Language models and how they work

• Large language model (LLM): A model trained on vast amounts of text to predict and generate language.
• Generative AI: AI that creates new content — text, images, audio, or code — rather than just classifying or predicting.
• Transformer: The neural network architecture behind most modern language models, built around attention.
• Attention: A mechanism that lets a model weigh which parts of the input matter most for each output.
• Token: A chunk of text — often part of a word — that a model reads and generates one at a time.
• Tokenization: The process of breaking text into tokens before a model processes it.
• Context window: The maximum amount of text a model can consider at once, measured in tokens.
• Inference: Running a trained model to produce an output, as opposed to training it.
• Embedding: A numerical representation of text or data that captures meaning so machines can compare it.
• Foundation model: A large, general-purpose model trained broadly, then adapted to many specific uses.
• Multimodal: A model that handles more than one type of input or output, such as text and images together.

Working with models

• Prompt: The instruction or question you give a model to get a response.
• Prompt engineering: The practice of writing prompts to get better, more reliable outputs.
• System prompt: Background instructions that set a model's role and rules before a conversation begins.
• Zero-shot: Asking a model to do a task with no examples in the prompt.
• Few-shot: Including a few examples in the prompt to guide the model's response.
• Chain of thought: Prompting a model to reason step by step, which can improve complex answers.
• Temperature: A setting that controls how random or creative a model's output is.
• Retrieval-augmented generation (RAG): Giving a model relevant documents at query time so it answers from real sources.
• Grounding: Connecting a model's answers to verified data to reduce made-up information.
• Hallucination: When a model produces confident output that is wrong or fabricated.

Capabilities and systems

• Agent: An AI system that can take actions and use tools to complete multi-step tasks, not just chat.
• Chatbot: A program that converses with users in natural language, increasingly powered by LLMs.
• API: A connection that lets software send data to and receive results from an AI model.
• Open-source model: A model whose code or weights are publicly available to use and modify.
• Closed model: A proprietary model accessed through a vendor's service rather than downloaded.
• Computer vision: AI that interprets images and video.
• Natural language processing (NLP): The field focused on getting computers to understand and generate human language.

Performance, safety, and infrastructure

• GPU: A specialized processor that performs the parallel math AI training and inference rely on.
• Benchmark: A standardized test used to compare models on a task.
• Bias: Systematic skew in a model's outputs, often inherited from its training data.
• Alignment: The effort to make AI systems behave in line with human intentions and values.
• Guardrails: Rules and filters that constrain what a model is allowed to do or say.

Bottom line

You do not need to memorize all 50 terms to work with AI — but knowing them turns confusing jargon into a usable mental model. The most load-bearing ideas are the simplest: models learn patterns from data, predict outputs token by token, and can be wrong with full confidence. Keep this glossary handy, and when a new term appears, place it against these basics rather than treating it as something entirely new.

Sources and further reading

Sources

Wikipedia: Glossary of artificial intelligence
en.wikipedia.org