AI Glossary: Guide to essential terms and concepts
Artificial Intelligence (AI) is an exciting and rapidly expanding field. Continue reading to learn foundational AI concepts, as well as terms relevant to data processing and AI applications.
By: Jacob Given, Edited by: Gabriela Pérez Jordán, Reviewed by: Jeff Le
Published: June 19, 2025
Foundational AI concepts
Artificial intelligence
Artificial intelligence (AI) refers to computer systems that can accomplish tasks that typically require human intelligence, including interpretive tasks like sentiment analysis and text summarization.
- Generative artificial intelligence: Generative artificial intelligence (GenAI) uses a predictive model trained on vast amounts of data to produce new content, such as written text, images, and even video and audio.
- Artificial general intelligence: Artificial general intelligence (AGI) refers to AI systems capable of matching or surpassing human performance levels. At the present moment, AGI is a hypothetical achievement.
Machine learning
Machine learning (ML) refers to the use of algorithms that enable computers to learn autonomously and make predictions about trends based on data sets. An ML program can improve its accuracy over time as it processes more data.
- Supervised Learning: During supervised learning, a system learns based on human input in the form of labeled data. By feeding the system labeled examples, it becomes better at predicting the outputs associated with certain inputs.
- Unsupervised Learning: Unsupervised learning refers to a process in which an ML system identifies patterns in unlabeled data. This type of learning allows a system to discover patterns without relying on instructions from humans.
- Reinforcement Learning: Reinforcement learning refers to the trial-and-error process of training a model with positive and negative feedback based on its predictive performance. As the model receives feedback, it corrects its predictive processes accordingly.
- Deep learning: Deep learning (DL) leverages neural networks to simulate the learning activity of the human brain. DL allows AI models to extract patterns from large amounts of unstructured data.
- Neural Networks: Neural networks use a structure of nodes that mimics the neural activity of the human brain. A neural network consists of an input layer, several layers of nodes, and an output layer, as well as one or more hidden layers.
Natural Language Processing
Natural language processing (NLP) refers to computers' ability to understand and communicate in human language. Chatbots, virtual assistants, and many search engines rely on NLP.
- Large Language Models: Large Language Models (LLMs) are AI systems trained on a massive amount of text data, enabling them to understand and generate natural language. LLMs serve as foundation models, which means they can accomplish tasks in a variety of domains.
- Multimodal Models: Multimodal AI models can process and interact with a variety of input and output types, including text, video, and audio. They can generate output in one mode (text, for example) based on input from another mode (audio or video, for example).
AI data processing concepts
Algorithm
An algorithm is a set of rules or processes followed to accomplish a certain task. Algorithms can assist in discovering patterns in data sets or even offer predictions or suggestions based on historical data.
Data Mining
Data mining refers to the process of gleaning insights from large data sets. Through data mining, analysts can uncover hidden patterns that would otherwise remain undiscovered.
Big Data
Big data refers to high-volume data sets collected at high velocity by organizations from various sources. From these large data sets, organizations can derive actionable insights to make informed business decisions.
Real-world applications and interfaces of AI
Chatbot
Chatbots simulate conversations in natural language. A user provides an input, also known as a prompt, and the model provides an output, or a response.
- Prompts: Prompts refer to the inputs users provide to query an AI model. More specific and detailed prompts generally achieve more satisfactory results than vague ones.
- Hallucinations: A hallucination refers to an inaccurate or nonsensical output provided by an AI model. Hallucinations are an inherent limitation for AI models. According to a 2023 New York Times report, AI hallucinates an estimated 27% of the time.
Agentic AI
Agentic AI is an AI system capable of making decisions and performing tasks autonomously, with limited supervision. Leveraging a digital ecosystem of LLMs, ML, and NLP, agentic AI is designed to act on behalf of users or systems without having to rely on continuous prompts or human intervention. Autonomous vehicles and personalized learning assistants are examples of agentic AI.
Computer Vision
Computer vision refers to an AI subfield that aims to create systems that can extract insights from visual inputs like images and video.
Internet of Things
The Internet of Things (IoT) refers to physical objects with network connectivity that can collect and share data. IoT includes so-called "smart" devices, like smart cars, smart homes, and smart speakers.
Robotics
Robotics refers to the study of programmable machines that can replace human work, also known as robots. Robots typically operate autonomously, requiring minimal intervention to accomplish a task.
Build AI skills on edX
As AI continues to evolve rapidly, there's no better time to learn the field's fundamental concepts. If you're interested in diving deeper into artificial intelligence, our online beginner AI courses and programs are a great place to start.