Question 1
What does AI stand for?
-
Artificial Intelligence
-
Automated Internet
-
Applied Interface
-
Array Instruction
Question 2
What are AI systems designed to do?
-
Perform tasks associated with human intelligence
-
Repeat fixed instructions without using data
-
Store information without processing it
-
Connect devices without analysing signals
Question 3
What is machine learning?
-
Finding patterns in data to make predictions
-
Following rules written for every possible input
-
Storing examples without analysing them
-
Labelling every future input by hand
Question 4
What is training data?
-
Examples used to teach a model
-
New examples reserved for testing
-
Predictions produced after inference
-
Instructions that draw the user interface
Question 5
What is a model in AI?
-
A learned pattern used for predictions
-
A collection of raw training examples
-
A single prediction produced by a system
-
A tool used to label examples manually
Question 6
What does inference mean in AI?
-
Using a trained model to make a prediction
-
Adjusting a model using training examples
-
Comparing predictions with correct test labels
-
Attaching correct labels to training examples
Question 7
What does NLP stand for?
-
Natural Language Processing
-
Network Logic Protocol
-
Numeric List Processing
-
Node Link Programming
Question 8
What is computer vision used for?
-
Interpreting images and video
-
Interpreting written language
-
Converting speech into text
-
Predicting numerical values
Question 9
What is a chatbot designed to do?
-
Hold conversations in natural language
-
Identify faces in images
-
Recommend content from preferences
-
Convert recorded speech to text
Question 10
What do recommendation systems do?
-
Suggest products or content
-
Assign data to named categories
-
Convert recorded speech to text
-
Identify faces in images
Question 11
What does classification do?
-
Assigns data to named categories
-
Predicts a numerical value
-
Groups similar data without labels
-
Creates new examples from a prompt
Question 12
What does speech recognition convert?
-
Spoken words to text
-
Text to pixels
-
Images to audio
-
Numbers to passwords
Question 13
What can AI bias cause?
-
Systematic errors for some groups
-
Random errors spread equally across groups
-
Higher accuracy for every tested group
-
More representative training examples
Question 14
In AI, what does accuracy measure?
-
How often predictions are correct
-
How fast internet is
-
How many files are stored
-
How many colours are used
Question 15
What does human oversight of AI mean?
-
People check important AI decisions
-
People supply examples for training
-
Programmers write the original code
-
Users enter data into the system
Question 16
What can generative AI do?
-
Create new content
-
Assign existing data to categories
-
Measure a model's accuracy
-
Check whether users gave consent
Question 17
Face unlock on a phone is an example of what?
-
Facial recognition
-
Database normalisation
-
Binary shifting
-
Spreadsheet lookup
Question 18
Which is an AI privacy concern?
-
Collecting personal data without consent
-
Producing unequal accuracy between groups
-
Hiding the reasons for a decision
-
Replacing tasks previously done by people
Question 19
What is an AI hallucination?
-
A plausible-sounding but false output
-
A correct answer copied from training data
-
A model update approved by a person
-
A prediction with a high accuracy score
Question 20
What should good training data be like?
-
Representative, varied and accurately labelled
-
Large but collected from one group
-
Varied but labelled inconsistently
-
Current but missing uncommon cases
Question 21
What can introduce bias into training data?
-
Under-representing one group
-
Separating training and test examples
-
Checking labels for consistency
-
Sampling across the relevant population
Question 22
Why should humans review important AI outputs?
-
To catch errors and unsafe decisions
-
To add the outputs to training data
-
To make the model calculate faster
-
To collect more personal information
Question 23
What is test data used for in machine learning?
-
Evaluate a trained model on unseen examples
-
Adjust the model's learned patterns
-
Provide labels for the training examples
-
Store predictions for the user interface
Question 24
What does labelled training data contain?
-
Examples paired with correct categories
-
Examples without any expected outputs
-
Predictions made after deployment
-
Instructions written for every input
Question 25
What does transparency mean for an AI system?
-
Its decision process can be understood
-
Its training data is publicly editable
-
Its predictions are accepted without review
-
Its source code runs without testing
Question 26
What is a prediction in machine learning?
-
An output based on learned patterns
-
A guaranteed future outcome
-
A rule written directly by a programmer
-
A copy of a training example
Question 27
Why should an AI system be tested with new data?
-
To check whether it works beyond its training examples
-
To increase the size of the training set
-
To measure file compression
-
To replace human review
Question 28
What is object detection designed to do?
-
Identify objects and their positions in images
-
Identify speakers and their words in recordings
-
Group text documents by writing style
-
Predict numerical values from a table
Question 29
Who should remain accountable for an important AI-assisted decision?
-
The people or organisation using the system
-
The model without any human owner
-
The training examples stored by the system
-
The computer hardware running the model
Question 30
Why might an AI recommendation be unsuitable?
-
It may be based on incomplete or biased data
-
It has been checked by an expert
-
It uses complete and current data
-
It matches the user’s stated needs