Lost in AI terminology? Everything you need to know

Artificial Intelligence (AI) is a transformative field that combines computer science, data, and specialized algorithms to enable machines to perform tasks that typically require human intelligence. Navigating this field requires understanding its unique terminology. This vocabulary aims to be a foundational guide to the key concepts and terms in AI.

I. Core Concepts & Foundations

• Artificial Intelligence (AI): The overarching field dedicated to creating machines that can perform tasks traditionally requiring human intelligence, such as learning, problem-solving, decision-making, perception, and understanding language.
• Machine Learning (ML): A subfield of AI that enables systems to learn from data without explicit programming. It focuses on algorithms that allow computers to improve performance on a task with experience.
• Deep Learning (DL): A subfield of Machine Learning that uses artificial neural networks with multiple layers (deep neural networks) to learn from vast amounts of data. It’s particularly effective for complex pattern recognition tasks like image and speech recognition.
• Neural Network (NN): A computational model inspired by the structure and function of the human brain. It consists of interconnected nodes (neurons) organized in layers, processing information through connections.
• Algorithm: A set of well-defined instructions or rules designed to solve a problem or perform a task. In AI, algorithms are the computational recipes for learning and decision-making.
• Data: The raw facts, figures, and statistics that AI models learn from. Quality and quantity of data are critical for AI performance.
• Model: The output of a machine learning algorithm after it has been trained on data. It encapsulates the patterns and relationships learned from the data and is used to make predictions or decisions on new, unseen data.
• Inference: The process of using a trained AI model to make predictions or decisions on new, unseen data.
• Training: The process of feeding data to an AI model and adjusting its internal parameters so it can learn patterns and improve its performance on a specific task.
• Feature: An individual measurable property or characteristic of a phenomenon being observed. In ML, features are the input variables used by the model.
• Label (Target/Output Variable): The correct answer or desired output that an AI model is trying to predict or learn during supervised learning.
• Hyperparameters: Parameters that are set before the training process begins and control the learning process itself (e.g., learning rate, number of layers in a neural network). They are not learned from the data.
• Overfitting: A phenomenon where a model learns the training data too well, including its noise and outliers, leading to poor performance on new, unseen data.
• Underfitting: A phenomenon where a model is too simple to capture the underlying patterns in the training data, resulting in poor performance on both training and new data.
• Bias (in ML): A systematic error in a model’s predictions, often due to assumptions made in the learning algorithm or an unrepresentative training dataset. Also refers to ethical bias in models reflecting societal prejudices.
• Variance (in ML): The amount by which the model’s performance would change if it were trained on different training data. High variance can lead to overfitting.
• Generalization: A model’s ability to perform well on new, unseen data after being trained on a specific dataset. A well-generalized model avoids overfitting.

II. Types of Machine Learning

• Supervised Learning: Training a model on labeled data, where the model learns to map input features to known output labels.
  • Classification: A supervised learning task where the model predicts a categorical label (e.g., spam/not-spam, dog/cat, disease/no-disease).
  • Regression: A supervised learning task where the model predicts a continuous numerical value (e.g., house prices, temperature, stock prices).
• Unsupervised Learning: Training a model on unlabeled data to find hidden patterns, structures, or relationships within the data without explicit guidance.
  • Clustering: An unsupervised learning task that groups similar data points together into clusters (e.g., customer segmentation).
  • Dimensionality Reduction: An unsupervised learning technique that reduces the number of features in a dataset while preserving essential information (e.g., PCA).
• Reinforcement Learning (RL): A type of ML where an “agent” learns to make decisions by interacting with an environment, receiving rewards for desired actions and penalties for undesirable ones, aiming to maximize cumulative reward.
• Semi-supervised Learning: A hybrid approach that uses a small amount of labeled data combined with a large amount of unlabeled data during training.
• Transfer Learning: A technique where a model trained on one task is re-purposed or fine-tuned for a second, related task. This often involves using pre-trained models.
• Active Learning: A type of ML where the algorithm can interactively query a user or other information source to obtain the desired outputs (labels) for new data points.

III. Neural Networks & Deep Learning Specifics

• Neuron (Node): The fundamental unit of a neural network, which receives input, processes it, and transmits an output to other neurons.
• Layer: A collection of neurons in a neural network that perform similar functions.
  • Input Layer: The first layer of a neural network that receives the raw input data.
  • Hidden Layer: Layers between the input and output layers where the majority of computation and pattern extraction occurs. Deep learning networks have multiple hidden layers.
  • Output Layer: The final layer of a neural network that produces the model’s prediction or result.
• Activation Function: A mathematical function applied to the output of a neuron to introduce non-linearity into the network, allowing it to learn complex patterns (e.g., ReLU, Sigmoid, Tanh).
• Weights: Numerical values associated with the connections between neurons, representing the strength or importance of that connection. Weights are adjusted during training.
• Bias (in NN): A value added to the weighted sum of inputs in a neuron, allowing the activation function to be shifted.
• Backpropagation: The core algorithm used to train neural networks, which calculates the gradient of the loss function with respect to the weights, allowing for efficient weight updates.
• Loss Function (Cost Function): A mathematical function that measures how well the model’s predictions align with the actual values. The goal of training is to minimize this function.
• Optimizer: An algorithm (e.g., Adam, SGD) that adjusts the model’s internal parameters (weights and biases) during training to minimize the loss function.
• Learning Rate: A hyperparameter that determines the step size at which the model’s weights are updated during training.
• Epoch: One complete pass of the entire training dataset through the neural network during training.
• Batch Size: The number of training examples utilized in one iteration of training before the model’s parameters are updated.
• Convolutional Neural Network (CNN): A type of deep neural network particularly effective for processing grid-like data such as images. Used in computer vision tasks.
• Recurrent Neural Network (RNN): A type of deep neural network designed to process sequential data (e.g., time series, natural language) by maintaining an internal state or memory of previous inputs.
• Long Short-Term Memory (LSTM): A specialized type of RNN capable of learning long-term dependencies, overcoming the vanishing gradient problem in traditional RNNs.
• Transformer: A neural network architecture that relies on a mechanism called “attention” to weigh the importance of different parts of the input data, highly effective for natural language processing.
• Generative Adversarial Network (GAN): A framework consisting of two neural networks, a generator and a discriminator, that compete against each other to generate realistic synthetic data (e.g., images).

IV. Natural Language Processing (NLP)

• Natural Language Processing (NLP): A subfield of AI focused on enabling computers to understand, interpret, and generate human language.
• Large Language Model (LLM): A type of AI model (often based on Transformer architecture) trained on vast amounts of text data, capable of understanding, generating, and translating human-like text (e.g., GPT-4, Llama).
• Token: A unit of text, which can be a word, part of a word, or a single character, used as input for language models.
• Embedding: A numerical representation of words, phrases, or other data points in a multi-dimensional space, where similar items are mapped closer together. Crucial for LLMs to understand meaning.
• Prompt: The input text or query given to a generative AI model (especially LLMs) to elicit a specific response.
• Generative AI: A type of AI that can create new content, such as text, images, audio, or video, that is similar to human-created content.
• Sentiment Analysis: An NLP task that determines the emotional tone or sentiment expressed in a piece of text (e.g., positive, negative, neutral).
• Named Entity Recognition (NER): An NLP task that identifies and classifies named entities (e.g., persons, organizations, locations) in text.
• Machine Translation: An NLP task that automatically translates text or speech from one natural language to another.
• Text Summarization: An NLP task that generates a concise and coherent summary of a longer text document.
• Chatbot: An AI program designed to simulate human conversation, either through text or voice, allowing interaction with users.

V. Computer Vision (CV)

• Computer Vision (CV): A field of AI that enables computers to “see” and interpret visual information from images and videos.
• Image Recognition: A CV task that identifies and labels objects or features within an image.
• Object Detection: A CV task that identifies the presence, location, and class of multiple objects within an image or video.
• Image Segmentation: A CV task that divides an image into distinct regions or segments, often corresponding to different objects or parts of objects.
• Facial Recognition: A CV application that identifies or verifies a person from a digital image or a video frame.

VI. AI Deployment & Infrastructure

• Cloud AI: AI services and models hosted and delivered over the internet via cloud computing platforms (e.g., AWS AI, Google Cloud AI, Azure AI).
• Edge AI: AI processing that happens directly on a local device (the “edge”) rather than in a centralized cloud server, enabling faster responses, lower latency, and enhanced privacy.
• API (Application Programming Interface): A set of rules and protocols that allows different software applications to communicate and interact with each other. Many AI models are accessible via APIs.
• GPU (Graphics Processing Unit): A specialized electronic circuit designed to rapidly manipulate and alter memory to accelerate the creation of images. GPUs are highly parallel and are essential for training deep learning models.
• TPU (Tensor Processing Unit): A custom-built AI accelerator developed by Google specifically for neural network machine learning, particularly with TensorFlow.
• Containerization (e.g., Docker): Packaging software code and its dependencies into a standard unit called a container, ensuring it runs consistently across different environments. Important for deploying AI models.
• MLOps (Machine Learning Operations): A set of practices that combines Machine Learning, DevOps, and Data Engineering to standardize and streamline the lifecycle of ML models, from experimentation to deployment and monitoring.

VII. Ethical & Societal Aspects

• AI Ethics: The field that examines the moral implications of AI technologies, focusing on principles like fairness, accountability, transparency, and privacy.
• Algorithmic Bias: Systematic and repeatable errors in a computer system that create unfair outcomes, such as favoring one arbitrary group over others.
• Fairness (in AI): The principle that AI systems should treat all individuals and groups equitably, without prejudice or discrimination.
• Transparency (in AI): The ability to understand how an AI system works, including its decision-making processes, data sources, and potential biases.
• Accountability (in AI): The capacity to identify who is responsible for the actions and impacts of an AI system, especially in cases of errors or harm.
• Explainable AI (XAI): A set of techniques and tools that help humans understand why an AI model made a particular decision or prediction.
• Privacy (in AI): Protecting personal and sensitive data used by AI systems from unauthorized access, use, or disclosure.

This vocabulary is a living document, as the field of AI is constantly evolving. As you delve deeper, you’ll encounter even more specialized terms, but this list provides a solid foundation for understanding the core concepts and discussions in Artificial Intelligence.