Science & technology · free trivia

Artificial intelligence quiz

AI trivia about language-model terminology, evaluating learning and the role of data. Test the ideas behind words that appear in discussions of neural networks.

Free Artificial intelligence quiz: answer questions, see concise explanations and compare your result with friends.

Try in your browserStart the Artificial intelligence quizThe game opens in Telegram. If you are in Russia and the link does not open, check whether Telegram is available on your network.What is covered? ↓
  • Completely free
  • Answer explanations
  • Solo and friend duels

Inside this quiz

What this quiz covers

For players interested in Artificial intelligence. Play on your own, or choose a duel or a group room in Telegram.

Try three questions

Examples from the current published question bank. Questions and answers may reveal story details. Game questions and difficulty can vary; this is a preview, not a complete quiz.

Medium

What does overfitting look like?

  1. Good on new data, poor on training data
  2. Equally good on both datasets
  3. Good on training data, worse on new data
  4. Poor on both training and new data
Show answer and explanation

Good on training data, worse on new data

An overfit model performs well on training examples but substantially worse on new ones. It has learned training-specific details that generalize poorly.

Source

Medium

What is a token for a language model?

  1. The number of trainable model parameters
  2. A vector representing a whole document’s meaning
  3. A text unit, such as part of a word
  4. A probability estimate for the next word
Show answer and explanation

A text unit, such as part of a word

A token can be a word, part of a word, a character, or another text fragment.

Source

Medium

Why does training-data quality matter for model fairness?

  1. Data biases can carry into predictions
  2. A large dataset is necessarily balanced
  3. Random shuffling removes sampling biases
  4. The network architecture fully determines fairness
Show answer and explanation

Data biases can carry into predictions

Systematic biases in data can be reflected in predictions.

Source

What to expect: Artificial intelligence quiz

The samples compare model performance across datasets, define a term and examine why data quality matters. They concern machine-learning concepts rather than the feature list of a particular chatbot. Compare the meanings of the choices: a familiar technical word may still refer to the wrong part of a model's operation.

How it works

Choose a format and start answering.

The question set, scoring and short answer explanations are already prepared.

  1. 1

    Choose browser or Telegram

    Try a solo quiz as a guest in your browser, or open Mindroom in Telegram. Guest progress and results are not saved to a profile. Sign in to save progress and play with friends.

  2. 2

    Check your topic

    The topic link carries Artificial intelligence into game setup. Complete any first-visit steps, then check the selected topic, difficulty and timer before starting. See the solo quiz guide for settings and answer reviews.

  3. 3

    Invite friends for a match

    Sign in, choose a friend duel or group room and select the same topic. A room holds up to 16 players: everyone plays individually, receives the same questions and scores for accuracy and speed. Read how to play trivia with friends before sharing your invitation.

Question coverage

Topics in this Artificial intelligence quiz

language modelsevaluating learningdata quality

Ready to test yourself?

Play the Artificial intelligence quiz online.

Try a solo quiz in your browser without signing in. For saved progress and friend matches, sign in or open Mindroom in Telegram.

Play in your browser Start the quizThe game opens in Telegram. If you are in Russia and the link does not open, check whether Telegram is available on your network.