🤖 Artificial Intelligence Practice Test — 5th Grade
20 questions from 20 artificial intelligence articles. Take it online for an instant score, or print it with the answer key.
Artificial Intelligence test — 5th Grade
A model was right on 80 out of 100 tests. What is its accuracy?
Why do AI researchers like to test ideas with games?
What are the nerve cells in the brain called?
What is automation?
What does a self-driving car use to “see”?
What is a common reason for bias in AI?
How does a language model write an answer?
What does computer vision do?
What does copyright give?
What is the job of data in training an AI?
What is data labeling?
What is a dataset?
What is a deepfake?
Who is responsible for using AI ethically?
What is special about generative AI?
What does generative AI do?
Who proposed a test in 1950 to check whether a machine behaves intelligently?
What did a language model learn to do during training?
What happens in each step of training a model?
What is a neural network made of?
Name: ____________________ Date: ____________
A model was right on 80 out of 100 tests. What is its accuracy?
- A. 20 percent
- B. 100 percent
- C. 80 percent
- D. 8 percent
Why do AI researchers like to test ideas with games?
- A. Because games have no rules
- B. Because games are the only thing computers do
- C. Because games have clear rules and a simple way to know who won
- D. Because games don’t need a computer
What are the nerve cells in the brain called?
- A. Pixels
- B. Algorithms
- C. Robots
- D. Neurons
What is automation?
- A. A painted car
- B. A machine or program doing work automatically
- C. A kind of game
- D. Working without any tools
What does a self-driving car use to “see”?
- A. A printed map
- B. The passengers’ eyes
- C. Cameras, radar and other sensors
- D. Only the rearview mirror
What is a common reason for bias in AI?
- A. A screen that’s too small
- B. One-sided training data
- C. A broken keyboard
- D. A computer deciding to hate someone
How does a language model write an answer?
- A. It copies a whole answer from one book
- B. It picks a random answer from a short list
- C. A person sits and types every answer for it
- D. It predicts, word by word, what fits next
What does computer vision do?
- A. Cleans the screen
- B. Gets information from pictures and videos
- C. Prints photos
- D. Fixes eyeglasses
What does copyright give?
- A. Everyone the right to copy everything
- B. Money to everyone who goes online
- C. The creator the right to decide who uses their work
- D. Permission to change any work
What is the job of data in training an AI?
- A. It’s the rules the programmer writes by hand
- B. It’s the examples the program learns patterns from
- C. It’s the screen where you see the result
- D. It’s the electricity that powers the computer
What is data labeling?
- A. Painting the screen
- B. Deleting old data
- C. Adding the correct answer to each example
- D. Putting a sticker on the computer
What is a dataset?
- A. A kind of battery
- B. An organized collection of lots of data of the same kind
- C. A storage room for old computers
- D. A card game
What is a deepfake?
- A. A reliable news website
- B. A computer game
- C. A fake video, picture or voice made with AI
- D. A kind of camera
Who is responsible for using AI ethically?
- A. Only the computer
- B. The people who build and use it
- C. Nobody
- D. The AI itself
What is special about generative AI?
- A. It creates new content like text and pictures
- B. It can only count
- C. It works with no data at all
- D. It only turns computers on and off
What does generative AI do?
- A. Creates new content like pictures and text
- B. Fixes broken computers
- C. Only deletes files
- D. Only recognizes faces
Who proposed a test in 1950 to check whether a machine behaves intelligently?
- A. Garry Kasparov
- B. Alan Turing
- C. Karel Čapek
- D. Lee Sedol
What did a language model learn to do during training?
- A. Remember every person in the world
- B. See like a human eye
- C. Guess the next word in a text
- D. Feel emotions
What happens in each step of training a model?
- A. The model turns off and on
- B. The model guesses, the mistake is measured, and it’s adjusted a little
- C. The model picks what to learn by itself
- D. All the data is deleted
What is a neural network made of?
- A. Layers of connected calculating units
- B. Only electric wires
- C. Printed numbers
- D. Real nerve cells
Answer key
- C. 80 percent
- C. Because games have clear rules and a simple way to know who won
- D. Neurons
- B. A machine or program doing work automatically
- C. Cameras, radar and other sensors
- B. One-sided training data
- D. It predicts, word by word, what fits next
- B. Gets information from pictures and videos
- C. The creator the right to decide who uses their work
- B. It’s the examples the program learns patterns from
- C. Adding the correct answer to each example
- B. An organized collection of lots of data of the same kind
- C. A fake video, picture or voice made with AI
- B. The people who build and use it
- A. It creates new content like text and pictures
- A. Creates new content like pictures and text
- B. Alan Turing
- C. Guess the next word in a text
- B. The model guesses, the mistake is measured, and it’s adjusted a little
- A. Layers of connected calculating units
Answer key
Show the answers and explanations
- C. 80 percent — 80 correct answers out of 100 is 80 percent. (Model Accuracy)
- C. Because games have clear rules and a simple way to know who won — In a game it’s easy to measure success and compare programs. (AI and Games)
- D. Neurons — The brain is made of billions of neurons connected to each other. (AI vs. the Human Brain)
- B. A machine or program doing work automatically — Automation is getting work done without a person doing every step. (Automation)
- C. Cameras, radar and other sensors — Many sensors collect information, and an AI system combines it. (Self-Driving Car)
- B. One-sided training data — The system learns from the data, and if the data isn’t varied, it reflects the imbalance. (AI Bias)
- D. It predicts, word by word, what fits next — At each step the model calculates which word fits best, adds it and moves on to the next one. (Chatbots and Language Models)
- B. Gets information from pictures and videos — It’s a field that teaches computers to recognize, locate and track things in images. (Computer Vision)
- C. The creator the right to decide who uses their work — The law gives the creator control over how their original work is used. (Copyright)
- B. It’s the examples the program learns patterns from — In machine learning, the program learns from many examples — the data — instead of a programmer writing every rule. (Data and Training: How AI Learns)
- C. Adding the correct answer to each example — A label is the correct answer the model learns from. (Data Labeling)
- B. An organized collection of lots of data of the same kind — A dataset gathers many examples in one place, in order. (Dataset)
- C. A fake video, picture or voice made with AI — A deepfake is a fake that looks or sounds like a real person. (Deepfake)
- B. The people who build and use it — AI has no conscience; the responsibility belongs to people. (AI Ethics)
- A. It creates new content like text and pictures — Generative AI builds new results based on what it learned and on your request. (Generative AI)
- A. Creates new content like pictures and text — Generative AI creates content that didn’t exist before, based on a request. (AI That Makes Pictures)
- B. Alan Turing — In 1950 Alan Turing wrote a paper that asked whether machines can think and proposed the game named after him. (The History of AI)
- C. Guess the next word in a text — A language model was trained on tons of text and learned to predict what comes after what. (Language Model)
- B. The model guesses, the mistake is measured, and it’s adjusted a little — Training is made of repeating: guess, measure the mistake, make a small fix. (Model Training)
- A. Layers of connected calculating units — A neural network is a math structure of layers of units that pass numbers to each other. (Neural Network)
Study before the test
- 🎯 Model Accuracy
- 🎮 AI and Games
- 🧠 AI vs. the Human Brain
- ⚙️ Automation
- 🚗 Self-Driving Car
- ⚖️ AI Bias
- 💬 Chatbots and Language Models
- 👁️ Computer Vision
- ©️ Copyright
- 🗂️ Data and Training: How AI Learns
- 🏷️ Data Labeling
- 🗃️ Dataset
- 🎭 Deepfake
- 🧭 AI Ethics
- 🎨 Generative AI
- 🎨 AI That Makes Pictures
- 📜 The History of AI
- 💬 Language Model
- 🏋️ Model Training
- 🕸️ Neural Network
What is on this artificial intelligence test?
20 multiple-choice questions taken from 20 artificial intelligence articles written for 5th grade readers.
Can I print it?
Yes. Use the print button: the test and the answer key come out on separate pages, no sign-up needed.