AI and machine learning terminology is expanding rapidly — new terms emerge with every major research paper and model release. This reference covers 200+ essential terms across Core ML, NLP, Generative AI, Computer Vision, MLOps, Ethics, Architecture, Training, and Evaluation domains.
No terms match your search. Try different keywords or clear filters.
How to Use This AI & ML Glossary
The AI field has developed rapidly, creating a dense vocabulary that can be overwhelming for newcomers and specialists alike. This glossary covers 200+ terms across 9 major domains — from foundational machine learning concepts to cutting-edge generative AI terminology.
Navigating by Domain
Use the category filter buttons to focus on your area of interest. Core ML covers foundational concepts like neural networks, gradient descent, and model evaluation that underpin all AI systems. Generative AI focuses on LLMs, prompt engineering, and modern language model concepts. MLOps covers deployment, monitoring, and production systems.
Navigating Alphabetically
If you encounter an unfamiliar term while reading papers or documentation, use the letter filter to jump directly to terms starting with that letter. This is faster than scrolling through the full list when you know the term you're looking for.
Understanding Term Relationships
Many AI terms build on each other in a hierarchy. For example: attention mechanism → transformer → BERT/GPT → LLM → fine-tuning/prompting. When encountering a new term, check the "Related Terms" section in the definition to understand the conceptual neighbors. This helps build a mental map rather than memorizing isolated definitions.
Key Term Categories for Practitioners
For ML engineers building production systems, focus on: Architecture, Training, Evaluation, and MLOps categories. For prompt engineers and LLM users, focus on: Generative AI and NLP categories. For AI ethics and safety practitioners, focus on: Ethics and the alignment-related terms in Generative AI. For computer vision practitioners: Computer Vision and Architecture categories.
Commonly Confused Term Pairs
Several term pairs are frequently confused: precision vs. recall (both are evaluation metrics but measure different things), overfitting vs. underfitting (opposite model capacity problems), fine-tuning vs. prompting (weight updating vs. inference-time guidance), temperature vs. top-p (different randomness controls), and embedding vs. token (representation vs. unit of text). Each is defined clearly in this glossary with examples.
FAQ
What is the difference between AI, ML, and deep learning?
AI (Artificial Intelligence) is the broad field of creating machines that perform intelligent tasks. ML (Machine Learning) is a subset of AI where systems learn from data rather than explicit programming. Deep Learning is a subset of ML using multi-layer neural networks. All deep learning is ML, and all ML is AI — they're nested categories.
What does LLM stand for in AI?
LLM stands for Large Language Model. LLMs are deep learning models trained on massive text datasets that can generate, summarize, translate, and reason about natural language. Examples include GPT-4, Claude 3, Gemini 1.5, and Llama 3. The 'large' refers to both the model size (billions of parameters) and the training data scale.
What is the difference between a token and a word?
Tokens are the chunks that LLMs split text into for processing. A token is roughly 3-4 characters or about 0.75 words in English. Common words like 'the' are a single token, while rare words like 'supercalifragilistic' might be split into multiple tokens. API pricing is typically per token, not per word.
What is hallucination in AI?
Hallucination is when an AI model generates confident-sounding text that is factually incorrect or completely fabricated. Unlike human errors, AI hallucinations don't stem from malicious intent — the model produces plausible-sounding text based on statistical patterns without verifying factual accuracy. RAG (Retrieval-Augmented Generation) is the most common mitigation technique.
What is fine-tuning vs prompting?
Prompting (or prompt engineering) adjusts the model's behavior at inference time by crafting the input — no model weights change. Fine-tuning updates the model's internal weights by training it further on domain-specific data. Prompting is faster and cheaper; fine-tuning produces more reliable domain adaptation but requires a training dataset and compute budget.
Is this glossary free?
Yes, completely free. Search and browse all 200+ terms without signing up or creating an account. All content runs in your browser.
Is my data private?
Yes. This is a purely static reference tool. No data is collected, stored, or sent to any server. Your searches stay entirely on your device.