i'm aaron. i analyze, i educate, and i build things
terms you'll run into using ai, not just vibecoding
the text you give an ai model to tell it what you want. it can be a question, an instruction, examples, or all of them at once.
what you type into the ai. the clearer and more specific you are, the better what comes back.
"tighten the prompt." → say more exactly what you want, and what you don't, so the ai stops guessing.
the unit an ai model reads and writes in. a token is a chunk of text, often a few characters or part of a word, not a whole word.
ai doesn't count words, it counts tokens, little pieces of words. usage limits and pricing are measured in them.
"that's about 1,000 tokens." → roughly 750 words of text going into or out of the model.
the most text a model can take in at once, measured in tokens, counting both your input and its reply.
the model's short-term memory. fit your stuff inside it and the model sees all of it; go over and the oldest parts fall off.
"it has a 200k context window." → it can keep around 150,000 words in view at one time.
when a model produces information that is false or made up while presenting it confidently as fact.
the ai making something up and sounding sure about it. it isn't lying, it's filling a gap with plausible-sounding text.
"double-check it, models hallucinate citations." → it might invent sources, quotes, or numbers that look real but aren't.