Artificial intelligence is no longer limited to the idea of robots interacting with humans. Today, AI technologies are used in search engines, recommendation systems, smartphones, image generation, translation tools, autonomous vehicles, healthcare, and software development.

As AI evolves rapidly, new concepts and terminology appear constantly. This glossary aims to provide simple and concise explanations for the most common artificial intelligence terms.

This article was originally created in 2019 and has been updated to reflect modern AI concepts and technologies.

Artificial Intelligence Terms

  • Artificial intelligence (AI): A broad field of computer science focused on creating systems capable of performing tasks that normally require human intelligence, such as reasoning, language understanding, pattern recognition, decision making, and problem solving.

  • Artificial general intelligence (AGI): A hypothetical form of AI capable of understanding, learning, and performing nearly any intellectual task at a human level or beyond. AGI does not currently exist.

  • Artificial narrow intelligence (ANI): Also known as weak AI. AI systems specialized for specific tasks such as translation, image recognition, recommendation systems, or playing games.

  • Artificial superintelligence (ASI): A theoretical form of intelligence that surpasses human intelligence in nearly every field. ASI remains speculative.

  • Agentic AI: AI systems capable of autonomously planning, reasoning, using tools, and completing multi-step tasks with limited human intervention.

  • Algorithm: A set of instructions or rules designed to solve a problem or complete a task.

  • Alignment: The process of ensuring AI systems behave according to human intentions, values, and safety expectations.

  • Artificial neural network (ANN): A computing system inspired by biological neural networks that learns patterns from data.

  • Backpropagation: A neural network training method where prediction errors are propagated backward through the network to improve accuracy.

  • Bayesian network: A probabilistic graphical model representing variables and their conditional dependencies.

  • Big data: Extremely large and complex datasets that require advanced tools and techniques to process and analyze.

  • Chatbot: Software designed to simulate human conversation using text or voice interaction.

  • Classification: A machine learning task where data is assigned to predefined categories.

  • Clustering: A machine learning method that groups similar data points without predefined labels.

  • Cognitive computing: Systems designed to simulate aspects of human reasoning and decision making.

  • Computer vision: A field of AI that enables machines to interpret and analyze images and videos.

  • Convolutional neural network (CNN): A neural network architecture commonly used for image recognition and visual processing tasks.

  • Data mining: The process of discovering useful patterns, relationships, and insights from large datasets.

  • Dataset: A structured collection of data used for training, testing, or evaluating machine learning models.

  • Deep learning: A branch of machine learning based on multi-layer neural networks capable of learning complex patterns.

  • Diffusion model: A type of generative AI model commonly used for image generation by gradually removing noise from random data.

  • Embeddings: Numerical vector representations of text, images, or other data that allow AI systems to understand semantic relationships.

  • Explainable AI (XAI): Techniques and methods that make AI system decisions easier for humans to understand.

  • Fine-tuning: The process of adapting a pre-trained AI model to perform better on a specialized task or dataset.

  • Foundation model: Large AI models trained on broad datasets that can be adapted to many downstream tasks.

  • Generative AI: AI systems capable of generating new content such as text, images, audio, video, or code.

  • Generative adversarial network (GAN): A type of neural network architecture consisting of competing generator and discriminator models used for creating realistic synthetic data.

  • Genetic algorithm: An optimization technique inspired by biological evolution and natural selection.

  • Hallucination: Incorrect or fabricated information confidently produced by an AI model.

  • Heuristic: A practical problem-solving approach designed to produce sufficiently good solutions efficiently.

  • Image recognition: The ability of AI systems to identify objects, people, scenes, or patterns within images.

  • Inference: The process of using a trained AI model to generate predictions or outputs from new data.

  • Large language model (LLM): A deep learning model trained on massive amounts of text data to understand and generate human-like language.

  • Limited memory AI: AI systems capable of temporarily using historical information to make decisions.

  • Machine learning (ML): A field of AI focused on algorithms that improve automatically through experience and data.

  • Machine translation: AI-powered translation of text or speech between languages.

  • Model: A mathematical system trained on data to make predictions or generate outputs.

  • Multimodal AI: AI systems capable of processing multiple forms of input simultaneously, such as text, images, audio, and video.

  • Natural language processing (NLP): A field of AI focused on enabling computers to understand, interpret, and generate human language.

  • Optical character recognition (OCR): Technology that converts printed or handwritten text from images into machine-readable text.

  • Overfitting: A machine learning problem where a model learns training data too specifically and performs poorly on new data.

  • Parameters: Internal numerical values learned by an AI model during training.

  • Pattern recognition: The automated detection of patterns and regularities in data.

  • Prompt: Instructions or input text given to an AI model.

  • Prompt engineering: The practice of designing prompts to achieve better AI outputs.

  • Reactive machines: Basic AI systems that react to current inputs without storing memories or learning from past experiences.

  • Reasoning model: An AI model optimized for logical analysis, step-by-step thinking, and problem solving.

  • Recurrent neural network (RNN): A neural network architecture designed for sequential data such as text, audio, or time-series information.

  • Reinforcement learning: A machine learning approach where agents learn by receiving rewards or penalties for actions.

  • Retrieval-augmented generation (RAG): A method combining AI generation with external knowledge retrieval to improve factual accuracy.

  • Robotics: The field involving the design, construction, and operation of robots.

  • Robotic process automation (RPA): Software automation technology used to perform repetitive digital tasks.

  • Structured data: Data organized into clearly defined formats such as tables and databases.

  • Supervised learning: Machine learning using labeled datasets where correct outputs are already known.

  • Synthetic data: Artificially generated data used for training or testing AI systems.

  • Token: A unit of text processed by language models. Tokens may represent words, subwords, or characters.

  • Training: The process of teaching an AI model using data.

  • Transfer learning: Reusing knowledge learned from one task to improve performance on another task.

  • Transformer: A neural network architecture introduced in 2017 that became the foundation for modern large language models.

  • Turing test: A test proposed by Alan Turing to evaluate whether a machine can imitate human conversation convincingly enough to fool humans.

  • Unstructured data: Data without a predefined format, such as audio, images, videos, and social media posts.

  • Unsupervised learning: Machine learning using unlabeled data to identify hidden patterns or structures.

  • Vector database: A database optimized for storing and searching embeddings used in semantic AI applications.

  • Zero-shot learning: The ability of an AI model to perform tasks it was not explicitly trained for.

Notes

  • Some concepts in AI evolve rapidly, so definitions may change over time.
  • Many modern AI applications combine multiple technologies such as transformers, multimodal processing, embeddings, and retrieval systems.
  • The rise of generative AI and large language models significantly expanded the public use of AI after 2022.