WorksheetsAI Foundations 1.1 to 1.8
Total questions: 40
Worksheet time: 21mins
Abstraction
Picking out the important parts and ignoring the rest
Teaching an AI model with lots of data
A network that uses weights to make decisions
The chance something will happen
A technology that mimics human thinking
Probability
Artificial intelligence (AI)
Hallucination
Prompt
Probability
The process of collecting data
The chance that an event will occur
A type of computer program
The study of shapes
Prompt
A network of connected nodes
A question or instruction given to an AI
Teaching an AI model
The chance of something happening
Which of these are "multimodal"?
Images and text
Video
Images
Text and Audio
Video and Images
An AI trained on a huge amount of information and capable of many things
Neural network
Large language model
Complex neural model
Robot
Neural Network
A set of instructions for AI
A collection of weighted nodes
Teaching an AI model
The chance of something happening
What does "training" mean for AI?
Teaching an AI to focus on important information
Giving an AI lots of data to help it learn
Adjusting the probability weights in a neural network
Asking an AI carefully crafted questions
What is "training data"?
The numeric weights assigned to a neural network
The information, like text or images, used to teach an AI model
The chance of something happening
Unexpected results from an AI while it is being trained.
Hallucination
AI responses that sound true but are not
AI responses are completely ridiculous
AI responses that are true, but unexpected
AI responses that are an opinion
Which is an example of abstraction in AI?
Ignoring the color of a car when figuring out its make and model
Figuring out the chance of rain tomorrow
Training a neural network with pictures
Giving a prompt to an AI
If you give an AI both text and pictures to use, what kind of model is it?
Large language model (LLM)
Multimodal model
Neural network
Hallucination model
A student wants to make an AI model more accurate. What should they work on?
Hallucination
Training
Prompting
Abstraction
Which situation shows the use of a prompt in AI?
Giving thousands of pictures to an AI model
Asking an AI, "What is the capital of France?"
Figuring out the chance of winning a game
Ignoring unimportant details in data
Why is training data important for AI?
It helps the model ignore unimportant details
It helps the model learn patterns and give better answers
It causes hallucinations that are harder to spot
It gives the AI a correct answer for every question
Why can a large language model (LLM) answer many different questions?
It is trained on a huge amount of information
It's memory contains every known topic
It has seen nearly every possible prompt before
It hasn't been trained and learns as it goes
A neural network is trained to read handwritten numbers. If it starts giving numbers that don’t exist, what is this called?
Probability
Hallucination
Prompting
Abstraction
You want to build an AI that can understand both spoken words and pictures. What kind of model should you use, and why?
Large language model
Multimodal model
Non-neural network
Hallucination model
If an AI gives wrong but believable answers, what will help?
Use more and better training data
Ignore the prompts given to the AI
Only use probability calculations
Remove all abstraction from the model
You want to make an AI that predicts the result of a coin toss. What idea would you use to figure out the chance of getting heads?
Abstraction
Probability
Hallucination
Training data
What is an "AI hallucination"?
When AI gives an answer that is not true or real
When AI answers very slowly
When AI only answers with numbers
When AI repeats the same answer
What does "AI bias" mean?
When AI always gives the same answer
When AI favors one type of information over another
When AI answers in different languages
When AI makes jokes
What is "context" in an AI prompt?
The color of the AI's screen
Background or extra information
The number of words in the answer
The time it takes to answer
What is "prompt engineering"?
Desigining better AI hardware
Writing prompts to get better AI responses
Drawing pictures for the AI
Fixing the AI's code
What does "alignment" mean for AI?
AI responds as wanted or expected
AI always says "yes"
AI only responds with facts
AI responds in the language and tone of the user
What is a contradiction in an AI's answer?
AI gives two answers that cannot both be true
AI gives a longer answer than needed
AI asks a question instead of giving an answer
AI says something the same as "I don't know"
If you ask the same question in two different ways and get two different answers from AI, what does this show?
AI is showing bias
Context is important in a prompt
AI is hallucinating
AI received new information
How can you get a better answer from AI when asking for a summary?
Give clear and detailed instructions
Limit the number of words
Tell the AI to improve its summary
Use simpler vocabulary
A researcher is designing an AI that avoids topics inappropriate for elementary students.
Which of these words most applies?
Alignment
Hallucination
Contradiction
Prompt engineering
What is the best action to take if you think an AI's response might be incorrect?
Verify the information from other reliable sources
Tell the AI it might be wrong and ask again
Use a different AI tool
Keep asking the AI until you get a different answer
Which of these is an example of AI bias?
AI gives different answers to one group of people than another
AI answers every question the same way
AI responds incorrectly to some math problems
AI makes a lot of mistakes
What should you do if you notice contradiction in AI responses?
Point out the problem and ask for a clear answer
Repeat the same prompt again
Always pick the second answer since it is likely correct
Stop using AI
Why is prompt engineering important?
It makes AI responses more useful
It helps AI respond faster
It changes the color of the AI's screen
It reduces spelling mistakes
If an AI gives an answer that does not match the facts, what is this called?
Hallucination
Alignment
Context
Prompt engineering
Which is the best way to avoid AI bias?
Use fair and balanced data to train the AI
Avoid controversial topics
Make the AI answer faster
Use only one type of question
If you want AI to answer in a certain way, what should you do?
Write a clear and specific prompt
Use a random question
Ask the same question many times
Use only yes or no questions
What does it mean if an AI is "aligned" with its users?
The responses match what they want or expect
It always responds with "yes" or "no"
It always responds factually to math and science questions
It always gives a complete response
A student asks AI for help with homework and gets an answer that is not true. What should the student do?
Check the answer with other sources
Always use the AI's answer
Ignore the homework
Ask a friend to do it
Which of these AI responses from the same conversation shows a contradiction?
"The sky is blue" "The sky is green"
"The sky is blue" "The sky is cloudy"
"The sky is blue"
"It is night"
"The sky is blue"
"The sky is pretty"
Why is context important for AI answers?
It helps AI give a better response
It helps AI respond more quickly
It changes language in which AI responds
It makes the AI response longer
