wayground logo

Free Printable Worksheets

Font size

S
M
L
XL
Worksheets

Pre-presentation quiz

Total questions: 19

Worksheet time: 10mins

Name
Class
Date
1.

What does "perturbation" mean in the context of adversarial examples?

a)

A large, random modification that changes the meaning of the input

b)

A small, intentional change made to an input to mislead a model

c)

A type of data used to improve model training

d)

A method to visualize model performance

2.

What is the main purpose of applying a perturbation to an input?

a)

To make the input more realistic

b)

To improve the accuracy of the model

c)

To make the model produce an incorrect output

d)

To reduce overfitting

3.

What are vulnerabilities in AI systems?

a)

Simple programming errors or syntax mistakes

b)

Structural weaknesses that cause incorrect or exploitable outputs

c)

Network latency problems in AI hardware

d)

Overfitting issues during training

4.

Why do overfitting, memorization, and brittleness make AI “vulnerable”?

a)

Because people don’t use Wi-Fi

b)

Because AI wants to rest

c)

Because the AI doesn’t actually understand — it just copies patterns

d)

Because AI hates learning

5.

Do AI vulnerabilities exist at the intersection of model architecture, training data quality, and deployment context?

a)

Yes

b)

No

6.

These are characteristics of LLM hallucinations except:

a)

Generating fluent and coherent text

b)

Producing factually incorrect or fabricated content

c)

Always caused by software or coding bugs

7.

Are deepfakes and hallucinations different because of human involvement?

a)

Yes

b)

No

8.

Deepfakes does not involve human manipulation

a)

Yes

b)

No

9.

Models purposely lie when they hallucinate, generating factually incorrect outputs through probabilistic mechanisms without intentional deception

a)

Yes

b)

No

10.

What is intrinsic hallucination?

a)

Output that is intentionally misleading by human manipulation

b)

Output that contains spelling or grammar errors

c)

Output that is fluent and coherent

d)

Output that contradicts the source content

11.

Faithfulness refers to when models stay consistent and truthful to the provided sources

a)

Yes

b)

No

12.

What is the acceptable tolerance level of hallucination in abstractive summarization?

a)

High

b)

Moderate

c)

Low

d)

Very low

13.

Hallucination in open-domain dialogue generation are generally unacceptable, unless they are minor and do not involve sever factual issues

a)

Yes

b)

No

14.

Hallucination in Data-to-Text occurs due to the gap between structured data and the natural language text produced during the conversion

a)

Yes

b)

No

15.

What factors contribute to LLM hallucination from data?

a)

Source-Reference Divergence

b)

Heuristic Data Collection

c)

Dataset Duplication

d)

Innate Divergence

16.

Hallucinations persist partly because current evaluation encourages guessing rather than honesty about uncertainty

a)

Yes

b)

No

17.

If an AI memorizes private information from its data, what could go wrong?

a)

It might share secrets by accident

b)

It becomes shy

c)

It learns faster

d)

It refuses to answer

18.

What are some real-world consequences of misinformation spreading online?

a)

Public confusion

b)

Political polarization

c)

Reputational harm

d)

Moral panic

19.

Why is AI-generated misinformation particularly dangerous compared to traditional misinformation?

a)

It spreads slower and is easier to detect

b)

It often looks and sounds more credible

c)

It can only be created by experts with coding skills

d)

It usually contains obvious factual errors