Quiz: What is Artificial Intelligence?

Question 1

Which statement is the best operational description of an AI system?

• ✓Software that uses a neural network
• ✓A machine-based system that infers outputs from inputs for objectives in a context
• ✓Any application marketed as intelligent
• ✓A system that learns continuously after deployment

Question 2

Which item belongs on the **authority** axis?

• ✓Large Language Model
• ✓Machine Learning
• ✓Generate text
• ✓Recommend but do not approve

Question 3

Can a symbolic expert system be AI without Machine Learning?

• ✓Yes, it can infer conclusions through explicit knowledge and rules
• ✓No, every AI system learns from data
• ✓Only if it generates text
• ✓Only if it adapts after deployment

Question 4

Why is `Generative AI` not a good value for the model-family axis?

• ✓It is a database feature
• ✓It describes an output capability that several model families can provide
• ✓It only applies to images
• ✓It is identical to an application

Question 5

An LLM drafts an insurance decision letter. What grants permission to send it?

• ✓The model's parameter count
• ✓A confident tone
• ✓The fact that generation succeeded
• ✓Application authorization, policy controls, and the approved workflow

Question 6

Which description is most precise for a gradient-boosted credit-risk service?

• ✓It is Generative AI because it creates a score
• ✓It is an AI application using Machine Learning to predict risk
• ✓It is a Foundation Model because several teams call it
• ✓It is an LLM because it supports a decision

Question 7

Which condition most strongly requires abstention from a policy assistant?

• ✓The response takes two seconds
• ✓The user asks a short question
• ✓Retrieved sources conflict and no precedence rule resolves them
• ✓The LLM produces grammatically correct text

Question 8

Why should document permissions be applied before retrieval context reaches the model?

• ✓It improves font rendering
• ✓It guarantees model accuracy
• ✓It prevents unauthorized content from entering model context, output, or traces
• ✓It removes the need for output validation

Question 9

Which statement about adaptation is correct?

• ✓Every intelligent system adapts online
• ✓A frozen model is not AI
• ✓Adaptation after deployment is optional and creates additional controls
• ✓Adaptation removes the need for versioning

Question 10

Which metric best tests whether a cited answer is grounded?

• ✓Response length
• ✓Model confidence alone
• ✓User-interface load time
• ✓Whether each material claim is entailed by an authorized cited source

Question 11

When is deterministic software preferable to AI?

• ✓When an authoritative formula or complete rule directly produces the answer
• ✓Whenever a problem involves text
• ✓Only when no data exists
• ✓Never, because AI is more flexible

Question 12

What makes an architect-level AI classification complete?

• ✓Naming the model provider
• ✓Drawing only the model endpoint
• ✓Connecting mechanism, evidence, authority, controls, failure paths, and ownership
• ✓Selecting the largest available model