AI Terminology Lab

🧠 AI Terminology Lab

Great! To learn and speak the language of AI confidently, it’s important to understand the core terminology. I’ll explain each one in simple, beginner-friendly terms — like talking to a smart 10-year-old — and group them into categories so it’s easy to remember.

🧠 Foundational Concepts

AI (Artificial Intelligence): Teaching machines to act smart, like humans.
ML (Machine Learning): Helping machines learn from data (not rules).
DL (Deep Learning): A type of ML using brain-like networks.
NLP (Natural Language Processing): How AI understands human language.
Computer Vision: How AI sees images or videos.
Model: The AI brain that makes decisions.

🧪 Machine Learning Terms

Training: Teaching a model with example data.
Dataset: A big collection of training data.
Features: Inputs the model learns from.
Labels: Correct answers in training data.
Supervised Learning: Learning with labeled data.
Unsupervised Learning: Finding patterns without labels.
Overfitting: When a model memorizes too much.
Underfitting: When a model is too simple.
Accuracy / Precision / Recall: Metrics to judge model quality.

🧠 Deep Learning & Neural Networks

Neuron: A tiny brain-like unit.
Layer: A group of neurons.
Neural Network: A system of layers that learns patterns.
CNN (Convolutional Neural Network): Good for image tasks.
RNN (Recurrent Neural Network): Good for sequences like text.
Transformer: A model type used in ChatGPT.
Embedding: Turning text/images into numbers.

🛠️ Tools & Systems

Prompt: What you type to an AI.
Agent: An AI that can plan and act.
Chain: A series of AI steps.
Inference: AI making a prediction.
API (Application Programming Interface): Way apps talk to AI.
LangChain: Toolkit to build smart AI apps.

🛡️ Ethics & Safety

Bias: When AI is unfair.
Fairness: Making AI treat people equally.
Explainability: Understanding AI decisions.
Privacy: Protecting personal info.
Responsible AI: Using AI carefully.

🧠 Advanced AI Terminology – Level 2

Term	Full Form	Simple Explanation
LLM	Large Language Model	A big AI trained to understand and generate human-like text (e.g. GPT-4).
RAG	Retrieval-Augmented Generation	A method where the AI fetches facts from documents before answering.
FAISS	Facebook AI Similarity Search	A tool to store and search text as vectors (used in RAG for memory).
CoT	Chain of Thought	Teaching AI to solve problems step-by-step, like showing its work.
ReAct	Reason + Act	AI thinks out loud and then takes an action (used in agents).
API	Application Programming Interface	How apps and tools communicate with AI services like OpenAI.
LangChain	–	A framework to build AI apps that use tools, memory, and agents.
Embedding	–	Turning words into numbers so they can be stored or searched.
Agent	–	An AI that can plan steps and use tools—like a digital assistant.

← Back to Prompt Portfolio