Hub Terminology Library

Common AI Terminology:

AI Agent: Unlike a standard chatbot that just answers questions, an AI agent is designed to take autonomous action. It can break down a complex goal, browse the web, use software tools, and execute a multi-step plan to achieve a result. 

Algorithm: A specific set of mathematical instructions or rules given to a computer program to help it analyze data, learn patterns, or solve problems. 

Artificial Intelligence (AI): The broad field of computer science focused on creating systems and machines capable of performing tasks that typically require human intelligence, such as reasoning, learning, and problem-solving. 

Artificial General Intelligence (AGI): A theoretical milestone where an AI possesses human-level cognitive abilities across all domains. Unlike today's "Narrow AI" (which excels at specific tasks like writing or playing chess), an AGI could learn to do essentially any intellectual task a human can do.

AI Alignment: The crucial field of research dedicated to ensuring that AI systems are designed so that their goals, behaviors, and ultimate outcomes remain safe and perfectly aligned with human values.

Deep Learning: A highly specialized subset of machine learning inspired by the structure of the human brain. It uses complex, multi-layered artificial neural networks to process vast amounts of unstructured data.

Fine-Tuning: The process of taking an existing, pre-trained AI model and giving it additional training on a smaller, highly specific dataset so it performs better on a specialized task.

Generative AI: A category of artificial intelligence designed to create new, original content—such as text, images, audio, video, or code—based on the patterns it learned during its training.

Hallucination: An instance where an AI model generates incorrect, fabricated, or nonsensical information but presents it confidently as if it were a proven fact.

Inference: The operational phase of AI where a fully trained model is put to work in the real world, analyzing new, unseen data to make predictions or generate responses.

Large Language Model (LLM): A powerful type of AI system trained on massive amounts of text data, enabling it to understand, translate, summarize, and generate human language with high fluency.

Machine Learning (ML): A subfield of AI where computer systems are designed to learn from data and improve their performance over time without being explicitly programmed for every single action.

Multimodal AI: An advanced AI system capable of understanding, processing, and combining multiple types of data inputs simultaneously, such as reading text, analyzing images, and listening to audio all at once.

Natural Language Processing (NLP): The branch of artificial intelligence that focuses on the interaction between computers and humans through language, giving machines the ability to interpret and manipulate speech and text.

Neural Network: A computational architecture used in machine learning that mimics the interconnected neurons in a biological brain, allowing the system to recognize complex patterns and relationships in data. 

Parameters: The internal variables or "weights" that an AI model adjusts and learns during training. You can think of them as the artificial synapses in a neural network. When you hear about an AI having "70 billion parameters," it refers to the size and complexity of its artificial brain. 

Prompt Engineering: The skill and practice of crafting, structuring, and refining the input text (prompts) given to a Generative AI model to achieve the most accurate and useful output possible.

Recursive Self Improvement (RSI): Recursive Self-Improvement refers to the theoretical ability of an artificial intelligence system to analyze its own architecture, identify inefficiencies, and rewrite its own source code or algorithms to make itself smarter and more capable—without human intervention. 

Token: The fundamental unit of data that a language model reads and processes. A token can represent a single character, a syllable, or an entire word depending on the language and the specific model.

Training Data: The massive initial dataset—often consisting of text, images, or numerical records—used to teach an AI model how to recognize patterns and make decisions before it is released for use.