Comprehensive AI dictionary – from basic concepts to advanced technologies.
An autonomous AI system working in a loop: plan, execute, review, fix.
Research ensuring AI systems act in accordance with human values.
An AI tool integrated into an editor that helps write and fix code.
Using AI to find and fix bugs in code.
Legal frameworks and rules for AI development and deployment.
An interface through which applications communicate and share data.
Hypothetical AI capable of solving any intellectual task like a human.
A mechanism allowing a model to focus on relevant parts of the input.
A standardized test for comparing AI model performance.
Systematic error in an AI model caused by imbalanced training data.
A prompt technique where the AI model explains its reasoning step by step.
A conversational AI interface enabling dialogue interaction with users.
An AI feature that runs code and analyzes data directly in chat.
A branch of AI focused on understanding visual data — images and video.
The maximum amount of text an AI model can process at once.
A type of neural network specialized for processing image data.
Graphical representation of data for easier understanding of trends and patterns.
A subset of machine learning using multi-layered neural networks.
In-depth AI research with iterative searching and citations.
AI-generated fake image, video, or voice mimicking a real person.
A generative model that creates images by gradually removing noise.
A numerical representation of text, images, or other data in vector space.
An AI system's ability to explain why it reached its decision.
A principle ensuring AI systems treat all groups equally.
A technique where an AI model solves tasks based on just a few examples.
Adapting a pre-trained AI model to specific data or tasks.
An architecture of two competing networks — a generator and a discriminator.
AI systems capable of creating new content — text, images, music, video.
A technique anchoring AI responses in factual, verifiable sources.
When an AI model generates convincing-sounding but factually incorrect information.
The process where a trained model generates outputs on new data.
A large-scale AI model trained on massive amounts of text for language understanding and generation.
A branch of AI where systems learn from data without explicit programming.
An architecture where only part of the network is activated for each input.
Compressing a large AI model into a smaller one while preserving most capabilities.
An AI system processing multiple data types — text, images, audio, video.
A computational model inspired by biological neurons in the human brain.
A branch of AI focused on interaction between computers and human language.
Building automations and applications without writing code.
AI models with open source code and weights available to the public.
Using AI to forecast future trends based on historical data.
The art and science of crafting input queries for maximum quality AI responses.
A security attack where a user manipulates an AI model with hidden instructions.
A technique combining data retrieval with text generation for more accurate answers.
Restructuring code without changing its behavior for better readability and maintenance.
A type of machine learning where an agent learns by interacting with an environment through trial and error.
A technique for fine-tuning AI models using human feedback.
Artificially generated data used to train AI models.
The process of splitting text into smaller units (tokens) for AI model processing.
Transferring knowledge learned on one task to solve another.
A neural network architecture based on the self-attention mechanism.
An event that starts an automated workflow.
A database optimized for storing and searching vector embeddings.
Building applications using natural language and AI agents.
A URL that receives data from an application in real time.
An automated sequence of steps triggered by an event.
An AI model's ability to solve tasks it was not specifically trained for.