Prompting AI, explained quiz

Question 1

What is a 'prompt' when working with an AI like a chatbot?

Accepted Answer

The text or instruction you give an AI to guide its response

Answer

The keyboard shortcut used to start an AI program

Answer

The computer code used to train an AI model

Answer

The final output or answer that an AI produces

Question 2

Why does being specific in your prompt help the AI give a better answer?

Accepted Answer

It reduces ambiguity so the AI can produce a more accurate and relevant answer

Answer

It makes the AI run faster on your device

Answer

It allows the AI to search the internet more accurately

Answer

It increases the AI's memory size temporarily

Question 3

What is 'zero-shot prompting'?

Accepted Answer

Asking the AI to perform a task without providing any examples beforehand

Answer

Prompting the AI using zero words

Answer

Giving the AI no instructions so it responds however it likes

Answer

Running the AI without an internet connection

Question 4

What is 'few-shot prompting' and why is it useful?

Accepted Answer

Providing a small number of examples to guide the AI's format and response style

Answer

Sending the AI only a few words as the entire prompt

Answer

Training the AI from scratch on a small dataset

Answer

Asking the AI only a few questions per session to save resources

Question 5

What is 'chain-of-thought' prompting?

Accepted Answer

Instructing the AI to reveal its reasoning step by step before giving the final answer

Answer

Sending a series of short prompts one after the other in quick succession

Answer

Asking the AI to search a chain of websites for information

Answer

Connecting multiple AI models together in a processing chain

Question 6

What is a 'system prompt' in an AI application?

Accepted Answer

A hidden instruction set at the start of a session that defines the AI's role and rules

Answer

A prompt typed by the user to start a new conversation

Answer

The operating system's instructions for running the AI program

Answer

A prompt stored permanently on the computer's hard drive

Question 7

Why might an AI give slightly different answers each time you send the exact same prompt?

Accepted Answer

Because a temperature setting adds controlled randomness to the AI's output each time

Answer

Because the AI searches a different website each time the prompt is sent

Answer

Because the AI's training data is updated daily

Answer

Because the prompt is internally translated to a different language each session

Question 8

What is AI 'hallucination'?

Accepted Answer

When an AI confidently states false or made-up information as if it were true

Answer

When an AI creates realistic images from text descriptions

Answer

When an AI takes too long to generate a response

Answer

When an AI refuses to answer a question due to safety rules

Question 9

What is 'prompt injection' and why is it a security concern?

Accepted Answer

A security attack where hidden instructions inside processed content hijack the AI's behavior

Answer

Typing a prompt very quickly to overwhelm and confuse the AI

Answer

Adding extra examples to a prompt to improve its performance

Answer

Injecting extra code directly into the AI's training data

Question 10

What does the 'temperature' setting control in an AI language model?

Accepted Answer

The level of randomness and creativity in the AI's word choices and output

Answer

How fast the AI's processor runs to generate text

Answer

The physical heat generated by the AI's server hardware

Answer

The number of words the AI is permitted to generate per response

Question 11

What is the 'context window' of an AI model, and why does it matter?

Accepted Answer

The maximum amount of text the AI can read and remember at one time

Answer

The size of the screen used to display the AI's response

Answer

The number of different topics the AI can discuss simultaneously

Answer

The background application window where the AI stores its training files

Question 12

How do examples inside a prompt help reduce mistakes in the AI's output?

Accepted Answer

They show the AI the expected format and style, reducing guesswork and mistakes

Answer

They speed up the AI's response generation time

Answer

They permanently add new knowledge to the AI's training

Answer

They allow the AI to bypass its built-in safety checks

Question 13

What is 'role prompting', and give an example of how it changes an AI's response?

Accepted Answer

Giving the AI a character or expert persona to adopt throughout its responses

Answer

Asking the AI to list the roles in a movie or play

Answer

Assigning different tasks to separate AI models at the same time

Answer

Choosing which AI feature to activate from a settings menu

Question 14

What is 'iterative prompting'?

Accepted Answer

Refining your prompt step by step based on reviewing the AI's previous responses

Answer

Sending the exact same prompt over and over without any changes

Answer

Gradually building a prompt by adding exactly one word at a time

Answer

Using a Python loop to automatically send the same prompt to the AI

Question 15

What is a 'negative prompt', and where is it commonly used?

Accepted Answer

An instruction that tells the AI what to avoid or exclude from its output

Answer

A prompt that is written in a rude or unfriendly tone

Answer

A prompt that asks the AI to criticize or find flaws in something

Answer

A prompt that causes the AI to give a pessimistic or negative answer

Question 16

What is the key difference between prompting an AI and training an AI?

Accepted Answer

Training teaches the AI using data before it is deployed; prompting guides an already-trained AI at use time

Answer

Prompting is faster but both methods permanently change the AI's knowledge

Answer

Training uses text prompts while prompting uses visual or audio data

Answer

They are the same thing — both involve giving the AI examples to learn new facts from

Question 17

Why should you always fact-check information produced by an AI before using it?

Accepted Answer

Because AI generates text based on word patterns, not verified truth, and can be confidently wrong

Answer

Because AI always uses information from exactly one year ago

Answer

Because AI is programmed to deliberately include one false fact per response

Answer

Because AI responses are automatically encrypted and hard to read accurately

Question 18

What is a 'token' in the context of AI language models?

Accepted Answer

A small chunk of text such as a word or word fragment that an AI processes as one unit

Answer

A digital coin used to pay for AI services online

Answer

A secret password required to access an AI model

Answer

A special command that unlocks advanced AI features

Question 19

What is 'prompt engineering' as a skill?

Accepted Answer

Designing and optimizing prompts to reliably get high-quality outputs from AI models

Answer

Writing the computer code that builds and trains AI models from scratch

Answer

Engineering the hardware servers that AI models run on

Answer

Creating and labeling the datasets used to train AI models

Question 20

Why does telling an AI the target audience for your request improve its response?

Accepted Answer

Because the AI adjusts its vocabulary and complexity to suit the specified audience

Answer

Because the AI searches for content written by that specific audience group

Answer

Because the AI will send the response directly to that audience's email address

Answer

Because the AI automatically shortens its response for younger audiences

Question 21

How does adding 'think step by step' to a math prompt improve the AI's accuracy?

Accepted Answer

It causes the AI to generate each reasoning step, catching errors before they compound

Answer

It tells the AI to switch to a built-in calculator instead of generating text

Answer

It slows the AI's processor so it can be more careful about each word

Answer

It makes the AI search the internet for the solution step by step

Question 22

What ethical issue arises when you prompt an AI to impersonate a real, living person?

Accepted Answer

The AI can produce false statements attributed to that person, spreading harmful misinformation

Answer

The AI begins to access and use that person's private personal data

Answer

The AI becomes permanently linked to that person's identity in its memory

Answer

The AI violates its own memory storage limits and crashes

Question 23

What tends to happen when a single prompt contains contradictory instructions?

Accepted Answer

The AI may produce a confused response or try to partially satisfy both conflicting instructions

Answer

The AI automatically picks the better instruction and ignores the other cleanly

Answer

The AI crashes and must be completely restarted

Answer

The AI always asks the user which of the two instructions to follow

Question 24

How does requesting a specific output format in your prompt improve the usefulness of the AI's response?

Accepted Answer

It instructs the AI how to arrange its output, making the result easier to read and use directly

Answer

It prevents the AI from using any punctuation or special characters in its response

Answer

It tells the AI which programming language to use when writing its reply

Answer

It reduces the total token count by automatically compressing the AI's response

Question 25

What is the risk of using a vague prompt when asking an AI to write code?

Accepted Answer

The AI may make wrong assumptions, miss edge cases, or solve a different problem than intended

Answer

The AI will refuse to generate any code until the prompt is longer

Answer

The AI will always generate code in an obscure or outdated programming language

Answer

The AI will always generate code that is far too long to be practical

Question 26

How does a persona instruction change the style of an AI's writing output?

Accepted Answer

It causes the AI to adopt the vocabulary, tone, and style of the specified character or persona

Answer

It makes the AI write in a different font style or color in the output

Answer

It restricts the AI to only discuss topics from that character's historical period

Answer

It switches which underlying AI model processes the request

Question 27

What is the purpose of using 'delimiters' such as triple backticks in a complex AI prompt?

Accepted Answer

To clearly separate sections of a prompt so the AI does not confuse instructions with data

Answer

To automatically add bullet points or numbers to the AI's output

Answer

To speed up the AI's processing by compressing the prompt before sending

Answer

To encrypt the prompt so only the AI model can read the contents

Question 28

Why can prompting alone not give an AI knowledge of events that happened after its training cutoff date?

Accepted Answer

Because the model's knowledge is fixed at its training cutoff and prompts cannot update its internal memory

Answer

Because prompts are limited to 100 words and cannot contain enough new information

Answer

Because the AI automatically deletes old information every time a new conversation begins

Answer

Because the AI needs special administrator permission to access any information after a certain date

Question 29

What is the difference between an 'open-ended' and a 'closed' prompt?

Accepted Answer

An open-ended prompt allows many possible response types; a closed prompt constrains format and content

Answer

An open-ended prompt works with an internet connection; a closed prompt works offline only

Answer

An open-ended prompt is always long; a closed prompt is always short

Answer

An open-ended prompt is used for generating images; a closed prompt is only for text

Question 30

How does adding a word-count limit to a prompt affect the quality of the AI's output?

Accepted Answer

It forces the AI to prioritize key information, though a limit that is too tight may cut important details

Answer

It always improves quality because shorter answers are universally better

Answer

It has no effect because AI models always ignore length instructions in prompts

Answer

It causes the AI to repeat content in order to meet the exact word count target

Test yourself