Glossary

AI that matches or exceeds human performance across all cognitive domains. No agreed definition. Timeline estimates range from 2 years to never-as-defined.

An AI system that plans, executes, and iterates on multi-step tasks with autonomy — rather than responding to a single prompt and waiting.

A technique that asks an AI model to work through a problem step by step before producing a final answer, improving accuracy on complex tasks.

The maximum amount of text an AI model can process in a single interaction. Larger windows allow longer documents and more complex instructions.

AI-generated synthetic media — video, audio, or images — designed to appear authentic. Used for both creative and malicious purposes.

Further training a pre-trained AI model on a specific dataset to improve performance on a particular task or domain.

When an AI model generates plausible-sounding but factually incorrect content — including fabricated citations and invented events — with full confidence.

An AI model trained on large amounts of text data to predict the next most likely token, producing fluent contextually appropriate responses.

Asking the AI to first generate an optimal prompt for a given task, then execute it — effectively directing the model to direct itself.

AI models that process and reason across multiple input types — text, images, audio, and video — rather than text alone.

Using the output of one prompt as the input to the next, enabling complex multi-step tasks a single prompt cannot reliably complete.

A curated, documented collection of tested, refined prompts for recurring professional use cases — a core asset of a compounding AI practice.

Enhancing AI responses by retrieving relevant information from a knowledge base before generating an answer, reducing hallucination.

A professional with one area of deep expertise combined with broad AI collaboration skills — the profile the AI era rewards most.

The unit of text AI models process — typically a word or sub-word. Most LLMs measure context window size and pricing in tokens.

A named, non-optional workflow step where specific factual claims and citations are checked against primary sources before output is used.