AI: Glossary of Terms

Machine Intelligence (AI)

Machine Intelligence, or AI, refers to the capability of machines to emulate tasks that generally necessitate human intelligence, such as learning, problem-solving, decision-making, and comprehending language. AI is realized by crafting algorithms and systems capable of processing, analyzing, and understanding vast amounts of data, and making decisions based on this data.

Interface for Programming Applications (API)

An API, short for Interface for Programming Applications, is a collection of protocols and rules that facilitate the interaction and data exchange between different software applications. It serves as a bridge, allowing disparate programs to collaborate and function together, regardless of the programming languages or technologies they were built with. APIs enable diverse software applications to communicate and share data, fostering a more integrated and smooth user experience.

Compute Unified Device Architecture (CUDA)

CUDA is a methodology that allows computers to tackle complex and large-scale problems by segmenting them into smaller parts and solving them concurrently. It enhances the computer’s performance and efficiency by utilizing specific components known as GPUs. The analogy would be having several friends assist you in solving a puzzle – it’s significantly quicker than attempting to solve it single-handedly.

The term “CUDA” is a proprietary trademark of NVIDIA Corporation, the company that pioneered this technology.

Data Manipulation

Data manipulation refers to the procedure of preparing raw data for incorporation into a machine learning model. This process includes tasks such as cleaning, transforming, and normalizing the data.

Deep Learning (DL)

Deep Learning, a subset of machine learning, employs deep neural networks with multiple layers to learn intricate patterns from data.


To enable a computer to comprehend language, words must be represented numerically, as computers can only understand numbers. An embedding is a technique used for this purpose. It involves converting a word, such as “cat”, into a numerical representation that encapsulates its meaning. This is achieved using a specific algorithm that considers the word in the context of surrounding words. The resulting number signifies the word’s meaning and can be used by the computer to understand the word’s significance and its relation to other words. For instance, the word “kitten” might have an embedding similar to “cat” due to their related meanings, while the word “dog” might have a different embedding from “cat” due to their distinct meanings. This enables the computer to understand the relationships between words and interpret language.

Feature Engineering

Feature engineering refers to the process of selecting and creating new attributes from raw data that can enhance the performance of a machine learning model.


The term “Freemium” is frequently used on this site. It simply denotes that the specific tool you’re viewing offers both free and premium options. Typically, the free tier offers limited but unrestricted usage of the tool, with additional access and features available in the premium tiers.

GAN (Generative Adversarial Network)

GANs are specialized computer programs that generate new entities, like images or music. They work by training two neural networks against each other, with one generating data and the other scrutinizing it. The iterative feedback loop between the two networks leads to the creation of highly realistic data.

Generative Art

This art form utilizes computer programs or algorithms to generate unique visual or audio outputs, often exploiting randomness or mathematical rules for unpredictable, occasionally chaotic outcomes.

GPT (Generative Pre-trained Transformer)

Developed by OpenAI, GPT is a language model proficient in generating natural-sounding text based on the context provided.

GLTR (Giant Language model Test Room)

GLTR is a tool designed to identify if text is written by humans or computers, through the analysis of word usage and probability.


GitHub is a collaborative platform dedicated to hosting software projects.

Google Colab

An online service, Google Colab, provides a platform for sharing and executing Python scripts in the cloud.

GPU (Graphics Processing Unit)

A GPU is a specialized chip adept at performing complex calculations required for rendering images and videos. With their ability to execute multiple computations swiftly, GPUs are ideal for tasks demanding substantial processing power.


LangChain is a tool that simplifies the connection of AI models to external data sources. This facilitates the creation of AI agents or chatbots capable of executing tasks on behalf of the user.

LLM (Large Language Model)

LLMs are machine learning models trained on extensive text data, allowing them to generate human-like text.

ML (Machine Learning)

ML represents a methodology enabling computers to learn from data, without explicit programming.

NLP (Natural Language Processing)

A facet of AI, NLP is dedicated to enabling machines to comprehend, process, and produce human language.

Neural Networks

These are machine learning algorithms modelled after the human brain structure and functionality.

Neural Radiance Fields (NeRF)

NeRF represents a category of deep learning models proficient in executing numerous tasks, such as creating images, spotting objects, and performing segmentation. The core inspiration behind NeRF is the utilization of a neural network to map the radiance present in an image, defined by the quantity of light emitted or reflected by an object.


OpenAI stands as a research entity devoted to cultivating and endorsing artificial intelligence technologies that uphold safety, transparency, and offer significant societal benefits.


Overfitting is a frequent issue in machine learning where the model exhibits excellent performance with the training data but struggles with new, unseen data. This problem usually arises when the model becomes overly complex and learns excessive details from the training data, leading to a lack of generalization.


A prompt serves as a text snippet used to initialize a large language model and steer its generation process.


Python is a renowned high-level programming language, celebrated for its ease, readability, and versatility, and it’s widely employed in AI tools.

Reinforcement Learning

Reinforcement Learning is a subset of machine learning where the model acquires knowledge through a process of trial and error, receiving either rewards or penalties for its actions, and subsequently adjusts its behavior.

Spatial Computing

Spatial computing involves the employment of technology to infuse digital data and experiences into the physical realm. This can encompass augmented reality – adding digital information to the real-world view, or virtual reality – enabling full immersion into a digital environment. Its applications are wide-ranging, including education, entertainment, and design, potentially transforming our interaction with the world.

Stable Diffusion

Stable Diffusion is responsible for producing intricate artistic images guided by text prompts. This open-source AI model, capable of image synthesis, is accessible to all. Installation of Stable Diffusion can be carried out locally using GitHub code, or through various online user interfaces that utilize Stable Diffusion models.

Supervised Learning

Supervised Learning is a form of machine learning where the training data comes with labels, and the model is instructed to predict based on the associations between the input data and corresponding labels.

Temporal Coherence

Temporal Coherence denotes the uniformity and continuity of data or patterns over time. This principle is crucial in domains such as computer vision, natural language processing, and time-series analysis, where AI models are required to process and comprehend evolving data.

In computer vision, temporal coherence can be related to the uniformity and steadiness of visual content in videos, where the continuity of objects and scenes is preserved across frames.

In natural language processing, it could indicate the uniformity and progression of information in a text or conversation, ensuring AI model generates logically coherent responses or summaries.

In time-series analysis, temporal coherence could be linked to the consistency of trends and patterns in the data, enabling the AI model to forecast future values based on past data.

Unsupervised Learning

Unsupervised Learning signifies a branch of machine learning where the training data is unlabeled, and the model is instructed to identify patterns and relationships within the data independently.


A webhook functions as a mechanism for one computer program to transmit a message or data to another program over the internet in real-time. This is achieved by sending the message or data to a designated URL, owned by the recipient program. Webhooks are frequently used to streamline processes and enhance the cooperation between different programs. They serve as an invaluable asset for developers seeking to design custom applications or establish integrations between various software systems.

Register New Account