AI Glossary: 25 Essential Terms (2026)

Artificial Intelligence (AI)

Computer systems that perform tasks typically requiring human intelligence — reasoning, learning, pattern recognition, language understanding.

Large Language Model (LLM)

AI models trained on massive text datasets to understand and generate human language. Examples: GPT-4, Claude, Gemini.

ChatGPT

OpenAI's conversational AI assistant launched in November 2022, built on the GPT family of language models.

Claude

Anthropic's AI assistant, known for its strong reasoning capabilities, constitutional AI training, and long context windows.

Gemini

Google's multimodal AI model family, capable of processing text, images, audio, and video.

Transformer

Neural network architecture introduced in 2017 ('Attention Is All You Need') that powers most modern language models.

Reinforcement Learning from Human Feedback (RLHF)

Training technique where models learn to align with human preferences through feedback loops.

Prompt Engineering

Crafting input prompts to get optimal outputs from language models. A key skill for working with AI.

Token

Basic unit of text that LLMs process. Roughly 4 characters or 3/4 of an English word.

Context Window

Maximum amount of text an LLM can consider at once. Modern models range from 8K to 1M+ tokens.

Multimodal AI

AI systems that process multiple types of data (text, image, audio, video) simultaneously.

Fine-tuning

Adapting a pre-trained model to specific tasks or domains by training on specialized data.

Hallucination

When an AI generates plausible-sounding but factually incorrect information.

Agent / Agentic AI

AI systems that can autonomously plan and execute multi-step tasks using tools.

RAG (Retrieval-Augmented Generation)

Technique where LLMs query external knowledge before generating responses, improving accuracy.

Embedding

Vector representation of text/data that captures semantic meaning for similarity search and ML tasks.

Mixture of Experts (MoE)

Architecture where different parts of a model specialize in different tasks, improving efficiency.

Open Source AI

AI models with publicly available weights and code. Examples: Llama, Mistral, Stable Diffusion.

AI Alignment

Research field ensuring AI systems pursue intended goals safely and beneficially.

AGI (Artificial General Intelligence)

Hypothetical AI matching human intelligence across all cognitive domains.

AI Safety

Field focused on preventing harmful AI behaviors, both near-term and existential risks.

Inference

Process of running a trained AI model to produce outputs on new inputs.

Training

Process of teaching an AI model by exposing it to large datasets and adjusting parameters.

Foundation Model

Large pre-trained model that serves as a base for many downstream applications.

Constitutional AI

Anthropic's training approach using principles to guide AI behavior without constant human feedback.

← Read latest AI news • Home →