AI Literacy Quiz: How Much Do You Really Know About AI?

Everyone uses AI chat now, but few people can say what a token is, why a model invents a citation, or what happens to the text they paste in. This AI literacy quiz draws ten questions from a pool of thirty covering tokens and context windows, training versus inference, hallucination, temperature, retrieval, fine-tuning, privacy, copyright, and more. Every answer comes with a short explanation, so you learn something whether you get it right or wrong.

Use it as a quick self-check, a warm-up for a class, or a way to settle an argument about what AI can and cannot do. The questions avoid anything that dates, so there are no version numbers or leaderboards, just the ideas that stay true from one release to the next. Take it again for a fresh set of AI quiz questions and answers, then send your score to the friend who still thinks the model is browsing the web.

Sample questions

1. When an AI model reads your message, what does it actually process?
   • Whole sentences as single units
   • Individual letters, one at a time
   • Tokens: chunks of text ranging from a few characters to a whole word
   • The dictionary definition of each word
2. Roughly how do tokens relate to English words?
   • One token is exactly one word
   • One token is one sentence
   • One token is one character
   • A token is about three-quarters of a word on average, so 1,000 tokens is roughly 750 words
3. What is a model's 'context window'?
   • How long the model takes to answer
   • The maximum amount of text it can consider at once, including the conversation and its reply
   • The number of topics it has been trained on
   • The time span its training data covers

Start the quiz above to see the answers and explanations. Each attempt draws 10 questions from a pool of 30.

The ideas every AI user should know

A large language model is a next-token predictor trained on an enormous amount of text. That single fact explains most of its quirks: it counts in tokens rather than letters, it can only consider what fits in its context window, it writes with the same confidence whether it is right or wrong, and it has no way to check the world unless the app around it provides a tool for search, code, or retrieval.

The practical consequences are simple. Verify names, numbers, and citations before you rely on them. Assume anything you paste is stored by the provider under its privacy policy, and strip identifiers you do not need. Remember that a knowledge cutoff means recent events are invisible without a tool, and that a benchmark score is a fixed exam, not a promise about your task.

Common AI myths this quiz busts

Statements people repeat about AI chat, and what is actually true:

• 'The model learns from my corrections.' It does not; only the current conversation changes.
• 'It looked that up online.' Only if the app gave it a browsing tool and it used one.
• 'It sounded sure, so it is probably right.' Confidence and accuracy are unrelated.
• 'AI detectors can prove who wrote something.' They estimate predictability and get it wrong both ways.
• 'Anything an AI writes is mine to copyright.' Rules vary by country and are still being settled.
• 'A bigger context window means it remembers everything.' Long chats still trim or lose early details.