Nielsen Norman Group published a practical AI glossary aimed at product and design practitioners. It covers core technical terms: tokens, context windows, agents, evals, and prompt injection, defined as they function in product and UX work, not as researchers or vendors use them.
The distinction matters. Vendor definitions drift toward marketing. Research definitions assume technical depth. This glossary anchors terms to how they behave in real product contexts, where the stakes are interface decisions and user mental models, not benchmark scores.
The full article is worth reading for the linked NN/G deep-dives attached to individual entries, and for the accompanying study guide on how AI works and how users think about it. If you are building vocabulary before a project kicks off, start there.
[READ ORIGINAL →]