This artificial intelligence glossary compiles the most important concepts used in the AI field. From neural networks to ethical concerns and machine learning techniques, this artificial intelligence glossary is your go-to reference to understand the fundamentals of modern AI.

AI (Artificial Intelligence): Field of computer science focused on creating systems capable of performing tasks that normally require human intelligence.

A

AI (Artificial Intelligence): Field of computer science focused on creating systems capable of performing tasks that normally require human intelligence.

API (Application Programming Interface): Set of definitions and protocols for building and integrating application software, allowing different systems to interact with each other.

Artificial General Intelligence (AGI): Type of AI that can understand, learn, and apply knowledge in a way similar to humans, with the capability to solve any cognitive problem.

B

Backpropagation: Method used in training neural networks, where errors are propagated backward through the network to adjust weights and improve accuracy.

Bot: Software program that performs automated tasks on the internet, including user interaction, order processing, and information provision.

C

Chatbot: AI program designed to simulate conversations with human users, often used in customer service, sales, and technical support.

Contextual Awareness: The ability of an AI system to understand and appropriately respond to the context of an interaction or query.

Conversational AI: Technology that allows devices to interact with human users naturally, using natural language for communication.

Computer Vision: Field of AI that trains computers to interpret and understand the visual world, identifying and processing images and videos.

D

Deep Learning: Subfield of machine learning that uses artificial neural networks with many layers (deep) to model and understand complex data.

Dialogue System: System designed to interact with humans through natural language, using techniques from natural language processing (NLP) and machine learning.

Domain-Specific AI: AI designed to perform specific tasks within a particular domain or area of expertise.

E

Edge Computing: Processing data near the source of data generation, reducing latency and bandwidth usage.

Ethical AI: Practice of developing and implementing AI in a way that respects ethical principles, including transparency, fairness, and responsibility.

F

Fine-Tuning: Process of adjusting a pre-trained AI model on a specific dataset to improve its performance on a particular task.

G

Generative Adversarial Network (GAN): Type of neural network consisting of two models (generator and discriminator) that train each other, used to generate new data from trained examples.

GPT (Generative Pre-trained Transformer): Language model developed by OpenAI, trained on large amounts of text to generate coherent and contextually relevant text.

H

Hyperparameter: Parameter whose values are set before the machine learning training process begins and used to control the training of the model.

I

Inference: Process of using a trained model to make predictions or decisions based on new data.

Intent Recognition: Ability of an AI system to identify the intention or purpose behind a user input, essential for chatbots and virtual assistants.

L

Language Model: AI model trained to predict the next word in a sequence of words, based on the context provided by previous words.

Large Language Model (LLM): Large-scale language model trained on vast amounts of textual data to understand and generate high-quality text.

Latent Space: Abstract representation of data in a lower-dimensional space, used in deep learning to capture hidden features.

M

Machine Learning: Subfield of AI that involves training algorithms to make predictions or decisions based on data.

Multimodal AI: AI that can process and integrate multiple forms of data, such as text, image, and audio, to provide richer and more contextualised responses.

N

Natural Language Processing (NLP): Subfield of AI focused on the interaction between computers and humans through natural language.

Neural Network: Set of algorithms designed to recognise patterns, modelling the way the human brain operates.

O

Overfitting: Situation where an AI model fits too closely to the training data, hindering its performance on new data.

OpenAI: AI research organisation that developed language models like GPT-3 and GPT-4.

P

Pre-trained Model: AI model that has been trained on a large dataset and can be fine-tuned for specific tasks, saving time and training resources.

Prompt: Input provided to a language model to generate a response, which can be a question, instruction, or context.

Proactive AI: AI that anticipates needs or issues and takes action to address or resolve them before the user requests.

R

Reinforcement Learning: Method of training AI where agents learn to take actions in an environment to maximise a cumulative reward.

Response Generation: Process of creating an appropriate response to a user input, essential for chatbots and virtual assistants.

Retrieval-Augmented Generation (RAG): An AI architecture that combines document retrieval with natural language generation.

S

Self-Supervised Learning: Machine learning technique where the model trains on unlabelled data parts, generating its own labels from the context.

Sentiment Analysis: NLP technique used to identify and extract opinions and emotions expressed in a text.

Supervised Learning: Method of training AI using labelled data, where the model learns to map inputs to correct outputs.

T

Tokenization: Process of dividing text into smaller units (tokens), such as words or sub-words, for processing by AI models.

Transfer Learning: Technique where a model trained on one task is reused and fine-tuned for another related task.

Transformer: Neural network architecture used in language models like GPT, which uses attention mechanisms to process and generate text.

U

Unsupervised Learning: Machine learning method where the model finds patterns and relationships in data without using predefined labels.

User Intent: Objective or purpose behind a user’s interaction with an AI system.

V

Voice Recognition: Technology that allows an AI system to recognise and process human speech.

W

Word Embedding: Vector representation of words that captures their meanings, semantics, and relationships with other words.

X

Explainable AI (XAI): Field of AI focused on creating systems whose decisions and actions can be understood and explained by humans.

Z

Zero-Shot Learning: Capability of an AI model to perform tasks for which it has not been explicitly trained, using knowledge from related tasks.