Useful AI Terms

There are a lot of terms floating around in the AI sphere. What follows is a list of some of the most useful. The distinction we have made between More Accessible and More Technical terms is somewhat arbitrary, but the idea is that you will come across the terms in the first list more often and in less technical settings than those in the second list.

More Accessible

  • AGI (Artificial General Intelligence): AI that can think like humans.

  • AI Agents: Autonomous programs that make decisions.

  • AI Alignment: Ensuring AI follows human values.

  • Hallucination: When AI generates false information.

  • AI Model: A trained AI system for a task.

  • Chatbot: An AI that simulates human conversation.

  • Context: Information AI retains for better responses.

  • Foundation Model: Large AI model adaptable to tasks.

  • Generative AI: AI that creates text, images, etc.

  • LLM (Large Language Model): AI trained on vast text data.

  • Machine Learning: AI improving from data experience.

  • NLP (Natural Language Processing): AI understanding human language.

  • Neural Network: AI model inspired by the brain.

  • Prompt Engineering: Crafting inputs to guide AI output.

  • RAG (Retrieval-Augmented Generation): AI combining search with responses.

More Technical

  • Compute: Processing power for AI models.

  • CoT (Chain of Thought): AI thinking step-by-step.

  • Deep Learning: AI learning through layered neural networks.

  • Embedding: Numeric representation of words for AI.

  • Explainability: How AI decisions are understood.

  • GPU: Hardware for fast AI processing.

  • Ground Truth: Verified data AI learns from.

  • Fine-tuning: Improving AI with specific training data.

  • Inference: AI making predictions from new data.

  • Parameters: AI's internal variables for learning.

  • Reasoning Model: AI that follows logical thinking.

  • Reinforcement Learning: AI learning from rewards and penalties.

  • Supervised Learning: AI trained on labeled data.

  • Tokenization: Breaking text into smaller parts.

  • Training: Teaching AI by adjusting its parameters.

  • Transformer: AI architecture for language processing.

  • Unsupervised Learning: AI finding patterns in unlabeled data.

  • Vibe Coding: AI-assisted coding via natural language prompts.

  • Weights: Values that shape AI learning.