Skip to content

Glossary (A - F)

Artificial Intelligence (AI): The simulation of human intelligence processes by machines, especially computer systems. (Beginner)

Artificial Narrow Intelligence (ANI): AI designed to perform one specific task well (e.g., facial recognition). (Beginner)

Artificial General Intelligence (AGI): A theoretical level of AI that can understand, learn, and apply its intelligence to any problem, much like a human. (Advanced)

Algorithm: A set of rules or instructions given to an AI to help it learn on its own. (Beginner)

Alignment: The process of ensuring that an AI system’s goals and behaviors are consistent with human values and intentions. (Advanced)

Application Programming Interface (API): A set of protocols that allows different software applications to communicate with each other, often used to integrate AI into existing business tools. (Intermediate)

Attention Mechanism: A component of neural networks that allows the model to focus on specific parts of the input data when making a prediction. (Advanced)

Augmented Intelligence: A design pattern for AI that emphasizes the human-centric partnership of people and AI working together. (Intermediate)

Agentic AI: A form of AI that moves beyond simple conversation to actually performing tasks, making decisions, and using tools autonomously to achieve a goal. (Advanced)

Autonomous Agents: AI programs that can act independently, manage their own tasks, and interact with other systems without constant human input. (Advanced)

Auto-GPT: An experimental open-source application showcasing the capabilities of the GPT-4 language model to act autonomously. (Advanced)

Abstraction: The process of simplifying complex systems by focusing on high-level patterns while hiding unnecessary internal details. (Intermediate)

Activation Function: A mathematical formula (like ReLU or Sigmoid) that determines if a specific “neuron” in a network should be activated or not. (Advanced)

Adaptive Learning: AI systems that adjust their content or pace in real-time based on an individual user’s performance or behavior. (Intermediate)

Adversarial Attack: Carefully designed inputs intended to trick an AI into making incorrect or malicious decisions. (Advanced)

AI Act (EU): The world’s first comprehensive legal framework for Artificial Intelligence, categorizing systems by risk levels. (Intermediate)

AI Safety: A field of research dedicated to ensuring that AI systems are reliable, predictable, and do not cause unintended harm. (Advanced)

Alternative Text (Alt Text): AI-generated descriptions for images that help visually impaired users and improve SEO accessibility. (Beginner)

Anomaly Detection: The identification of rare events or items in a dataset that differ significantly from the norm (used in fraud detection). (Intermediate)

Artificial Super Intelligence (ASI): A theoretical level of AI that would surpass human intelligence across every possible field and skill. (Advanced)

Augmented Reality (AR) AI: The use of AI to dynamically overlay digital information onto a user’s view of the real world. (Intermediate)

Autoencoder: A type of neural network used to learn efficient, compressed versions of data in an unsupervised way. (Advanced)

Big Data: Large, complex data sets that require advanced AI tools to analyze for patterns and trends. (Beginner)

Bias: Prejudice in favor of or against one thing, person, or group in a way considered to be unfair, which can be unintentionally learned by AI. (Intermediate)

Backpropagation: The primary algorithm used to train neural networks by calculating the gradient of the loss function. (Advanced)

BARD: Now Gemini, Google’s conversational generative artificial intelligence chatbot. (Beginner)

Big Model: A colloquial term for state-of-the-art foundation models or LLMs trained on massive datasets. (Beginner)

Black Box: An AI system whose internal decision-making process is hidden or too complex for humans to easily understand. (Intermediate)

Bounding Box: A virtual rectangle drawn by an AI around an object it has identified within an image. (Beginner)

Chatbot: A computer program designed to simulate conversation with human users. (Beginner)

Computer Vision: A field of AI that enables computers to interpret and understand visual information from the world (images/video). (Intermediate)

Context Window: The maximum amount of information (tokens) an AI can process and “remember” at one time during a conversation. (Intermediate)

Conversational AI: AI that can engage in human-like conversation, such as ChatGPT or Claude. (Beginner)

Constitutional AI: A method developed by Anthropic to train AI to be helpful, harmless, and honest based on a predefined set of principles. (Advanced)

Chain-of-Thought (CoT): A prompting technique that encourages the AI to explain its reasoning step-by-step before providing a final answer. (Intermediate)

Compute: The processing power (CPUs/GPUs) required to train and run AI models. (Intermediate)

Deep Learning: A subset of machine learning based on artificial neural networks with many layers (hence “deep”). (Intermediate)

Data Mining: The process of discovering patterns and knowledge from large amounts of data. (Intermediate)

Data Sovereignty: The concept that data is subject to the laws and governance structures of the country where it is collected. (Advanced)

Deepfake: Synthetic media in which a person in an existing image or video is replaced with someone else’s likeness using AI. (Intermediate)

Diffusion Model: A class of generative AI often used for image generation (like Midjourney or DALL-E). (Advanced)

Expert Systems: Early AI systems designed to emulate the decision-making ability of a human expert. (Beginner)

Edge AI: Running AI algorithms locally on a device (like a phone or camera) rather than in the cloud. (Advanced)

Ethical AI: The field of study and practice focused on making sure AI is developed and used responsibly. (Intermediate)

Embeddings: A way to represent words or concepts as numbers so that an AI can understand the relationships between them. (Advanced)

Fine-Tuning: The process of taking a pre-trained AI model and training it further on a specific dataset to improve its performance on a niche task. (Intermediate)

Foundation Model: A large AI model trained on a vast amount of data that can be adapted to a wide range of downstream tasks. (Intermediate)

Few-Shot Prompting: A technique where you provide the AI with a few examples of the task you want it to perform within the prompt. (Intermediate)

Failure Mode: A specific way in which an AI system can fail to perform its intended function or produce harmful results. (Advanced)