Have a Look at our AI Glossary
A structured AI glossary created by the AI Competence Academy to support AI literacy, responsible use,
and informed decision-making.
AI Glossary
A set of rules or instructions a computer follows to solve a problem or make a decision.
Technology that enables machines to perform tasks that normally require human intelligence, such as understanding language or recognizing images.
Systematic errors in AI outputs caused by biased data, design, or assumptions.
Principles guiding responsible, fair, and transparent use of artificial intelligence.
Very large datasets used to train AI systems.
An AI system whose internal decision-making process is not easily understandable.
When AI systems reinforce or increase existing social biases.
An automated program that performs tasks online, such as posting or replying.
An AI system that communicates with humans using text or voice.
An AI task that assigns data into predefined categories.
AI technology that enables machines to understand images and videos.
The ability of AI to adapt responses based on situational information.
Labeling data so AI systems can learn from it.
When real-world data changes over time, reducing AI accuracy.
AI-generated media (images, audio, video) that realistically imitates real people.
A subset of machine learning using layered neural networks.
Online environments where people are mainly exposed to opinions that reinforce their own beliefs.
AI that runs directly on devices rather than in the cloud.
A numerical representation of text or images that AI can process.
AI systems designed to make their decisions understandable to humans.
A situation where algorithms show users only content aligned with their existing views.
A large AI model trained on massive datasets and adapted for many tasks.
Adjusting a pre-trained AI model for a specific task.
Ensuring AI outcomes do not unfairly disadvantage certain groups.
AI that creates new content such as text, images, or music.
Ensuring AI outputs are based on factual, verifiable information rather than speculation.
Hardware commonly used to train and run AI models efficiently.
Rules and constraints that limit unsafe or unwanted AI behavior.
When AI generates information that is false or made up but sounds convincing.
A system where humans supervise or intervene in AI decisions.
AI systems combining multiple techniques or models.
Settings that control how an AI model learns.
The process of using a trained AI model to make predictions.
Combining AI with automation to improve workflows.
Information provided to an AI system for processing.
How easily humans can understand AI decisions.
Attempts to bypass safety restrictions in AI systems.
AI systems that learn or adapt during real-time use.
A structured database that connects information logically.
A simple machine learning method for classification and prediction.
AI trained on vast amounts of text to understand and generate language.
The time it takes an AI system to respond.
How fast an AI model updates during training.
A tag assigned to data for training purposes.
A subset of AI where systems learn patterns from data.
Teaching an AI system using example data.
AI that processes multiple data types (text, image, audio).
When AI performance degrades due to changing conditions.
AI techniques for understanding human language.
A computing system inspired by the human brain.
Irrelevant or misleading data that reduces AI accuracy.
Preparing data so AI can process it consistently.
When AI performs well on training data but poorly in real life.
AI models whose code is publicly available.
Improving AI performance and efficiency.
The result produced by an AI system.
An instruction or input given to an AI system.
Designing prompts to get better AI results.
An AI-generated estimate of future or unknown outcomes.
Building AI systems with data protection as a core principle.
Testing AI systems for accuracy and reliability.
A request for information from an AI system.
AI learning through rewards and penalties.
AI developed and used ethically and transparently.
Combining AI generation with external knowledge sources.
AI’s ability to perform reliably under varied conditions.
Artificially generated data used for training AI.
AI that detects emotions in text.
Training AI using labeled data.
AI’s ability to grow with increased demand.
Data used to teach an AI model.
Reusing knowledge from one AI task for another.
Organized groups using bots and AI to spread disinformation.
Clear communication about how AI works and is used.
AI learning patterns without labeled data.
Confidence that AI behaves reliably and ethically.
A practical application of AI technology.
Data used to test AI during training.
A system for storing AI embeddings.
AI that understands and generates spoken language.
AI designed for specific tasks, not general intelligence.
Using AI to streamline processes.
Marking AI-generated content for identification.
AI designed to be understandable and transparent.
Structured data format sometimes used in AI systems.
Using AI to improve efficiency and output.
AI performing tasks without specific training examples.
Security model assuming no implicit trust.
A unit describing extremely large data volumes.