Glossary (S - Z)
Sentiment Analysis: The use of AI to determine the emotional tone behind a piece of text (e.g., positive, negative, or neutral). (Intermediate)
Structured Data: Highly organized data that fits nicely into tables or databases (e.g., Excel spreadsheets). (Beginner)
Sentiment Analysis: See Sentiment Analysis. (Intermediate)
Strong AI: See Artificial General Intelligence (AGI). (Advanced)
Supervised Learning: Training an AI model where the data is already labeled with the “correct” answers. (Intermediate)
Semantic Search: A search method that understands the meaning and intent behind a query rather than just matching keywords. (Intermediate)
Scalability: The capability of an AI system to handle growing amounts of work or its potential to be enlarged to accommodate that growth. (Beginner)
Self-Supervised Learning: A type of training where the model creates its own labels from the input data (e.g., hiding a word in a sentence and trying to guess it). (Advanced)
Simulation: The use of AI to create a virtual model of a real-world process or system to test scenarios without real-world risk. (Intermediate)
Slow AI: Systems designed to use more compute time to “think” through a problem before answering (e.g., the Reasoning models like o1). (Advanced)
Softmax: A mathematical function often used at the final layer of a neural network to turn raw scores into probability percentages. (Advanced)
Sparks of AGI: A term used to describe when a large language model shows early, unexpected signs of general reasoning across multiple domains. (Advanced)
System 1 Thinking: In AI, describes fast, instinctive, and heuristic-based responses (the standard LLM approach). (Intermediate)
System 2 Thinking: In AI, describes slow, logical, and deliberative “step-by-step” reasoning (the Chain-of-Thought approach). (Intermediate)
Semantic Chunking: The process of breaking down large documents into smaller, meaningful pieces based on their context rather than just character count. (Advanced)
Self-Attention: The mechanism that allows a Transformer model to weigh the importance of different words in a sentence relative to each other. (Advanced)
Stochastic: Randomly determined; describes the non-deterministic nature of AI outputs where the system “rolls dice” to pick the next token. (Advanced)
Synthetic Data: Data that is artificially generated by an AI rather than collected from real-world events. (Advanced)
Transformer: The fundamental neural network architecture that powers modern AIs like ChatGPT and Gemini. (Intermediate)
Token: A unit of text that an AI model processes. One token is roughly 4 characters or 0.75 words. (Intermediate)
Temporal Data: Data that is specifically indexed by time, such as stock market prices, sensor logs, or historical weather patterns. (Intermediate)
Tensor: A multidimensional array of numbers used in deep learning to represent and process complex data. (Advanced)
Token Limit: The maximum number of tokens an AI can “pay attention to” at once before it starts forgetting earlier parts of the conversation. (Beginner)
Top-P (Nucleus Sampling): A technique used to control the diversity and “creativity” of an AI’s output by only considering the most likely tokens. (Advanced)
Tokenization: The process of splitting text into small units (tokens) that a computer can process. (Intermediate)
Training Data: The massive dataset used to teach an AI model its patterns and knowledge. (Beginner)
Turing Test: A test of a machine’s ability to exhibit intelligent behavior indistinguishable from that of a human. (Beginner)
Temperature: A setting that controls the “creativity” or randomness of an AI’s output. (Intermediate)
Unstructured Data: Data that doesn’t have a pre-defined format, such as emails, PDFs, images, and video. (Beginner)
Unsupervised Learning: Training an AI on data where the answers aren’t provided, forcing the system to find its own patterns. (Advanced)
User Intent: What a user is actually trying to achieve when they type a query or prompt. (Beginner)
Vector Database: A specialized database designed to store and search “embeddings” (numerical representations) efficiently. (Advanced)
Vector Space: A multi-dimensional space where embeddings are mapped; the distance between points represents how similar concepts are. (Advanced)
Voice Recognition: See Speech Recognition. (Beginner)
Vision-Language Model (VLM): An AI that can specifically understand and describe the relationship between images and text. (Advanced)
U-Net: A specific neural network architecture widely used in image segmentation and medical imaging. (Advanced)
Weak AI: See Artificial Narrow Intelligence (ANI). (Beginner)
Weights: Numerical values within a neural network that determine the strength of connections between “neurons.” (Advanced)
XAI (Explainable AI): Methods and techniques in the application of AI such that the results of the solution can be understood by human experts. (Intermediate)
Zero-Shot Prompting: Asking an AI to perform a task without giving it any prior examples. (Intermediate)
Value Alignment: The ongoing challenge of ensuring that an AI system’s goals and behaviors remain consistent with human ethics and intent. (Advanced)