1
WGU D685 - PRACTICAL APPLICATIONS OF PROMPT
ENGINEERING OBJECTIVE ASSESSMENT FINAL — NEWEST
2026/2027 ACTUAL EXAM 200 COMPLEX EXAM
QUESTIONS AND CORRECT VERIFIED ANSWERS |
ALREADY GRADED A+ (MOST RECENT!!)
1. A healthcare organization is implementing an AI system to analyze patient
records. The system performs well on the training data but fails to generalize
to new patient populations. Which fundamental limitation of AI does this
scenario best illustrate?
A) Lack of emotional intelligence
B) Dependence on training data
C) Limited computational power
D) Excessive algorithmic complexity
Answer: B) Dependence on training data
Rationale: AI systems are fundamentally dependent on their training data.
When the training data does not represent the broader population, the model
fails to generalize. This is a core limitation of AI—it cannot reason beyond the
patterns it has learned from its training data.
, 2
2. A marketing team uses an AI chatbot to handle customer inquiries. The
chatbot correctly identifies that a customer is asking about a product return
but suggests a return policy for a completely different product. This error is
best classified as:
A) True negative
B) True positive
C) False positive
D) False negative
Answer: C) False positive
Rationale: A false positive occurs when the model incorrectly generates
outputs that are not relevant or accurate to the input or task. The chatbot
correctly identified the intent (return) but provided incorrect information
(wrong product's policy), making it a false positive.
3. A data scientist is training a model to classify customer emails. She provides
500 labeled examples of spam and non-spam emails. This is an example of
which learning paradigm?
A) Unsupervised learning
B) Reinforcement learning
C) Supervised learning
D) Semi-supervised learning
Answer: C) Supervised learning
Rationale: Supervised learning uses labeled data where the model learns to
map inputs to outputs based on examples. The data scientist provided labeled
examples (spam/non-spam), which is the defining characteristic of supervised
learning.
, 3
4. A researcher wants to use an LLM to solve a complex math problem. She
asks the model to "show all your work and explain each step." This prompt
engineering technique is called:
A) Zero-shot prompting
B) Few-shot prompting
C) Chain-of-thought prompting
D) Self-consistency prompting
Answer: C) Chain-of-thought prompting
Rationale: Chain-of-thought (CoT) prompting encourages LLMs to explain their
reasoning and reveal intermediate steps. By asking the model to "show all
your work," the researcher is using CoT to improve the model's reasoning and
accuracy.
5. A company is developing an AI recruitment tool. The tool is trained on 10
years of hiring data from a male-dominated industry. Which type of bias is
most likely to affect this system?
A) Measurement bias
B) Sampling bias
C) Algorithmic bias
D) Confirmation bias
Answer: C) Algorithmic bias
Rationale: Algorithmic bias arises from assumptions in AI design that can
amplify existing data biases. The historical hiring data likely contains gender
bias, and the algorithm may learn and perpetuate these patterns, favoring
male candidates.
, 4
6. A user is interacting with an AI image generation tool. They start with "a
cat" and then refine to "a fluffy orange tabby cat sitting on a windowsill at
sunset." This process is best described as:
A) Zero-shot prompting
B) Iterative prompting
C) Chain-of-thought prompting
D) Few-shot prompting
Answer: B) Iterative prompting
Rationale: Iterative prompting involves refining prompts through successive
iterations by adding additional details, modifiers, or parameters to improve
the quality or specificity of the generated images. The user starts simple and
progressively adds detail.
7. An AI system is being used to diagnose diseases. A patient's rare condition
is missed because the training data contained very few examples of that
condition. This is an example of:
A) Algorithmic bias
B) Sampling bias
C) Measurement bias
D) Confirmation bias
WGU D685 - PRACTICAL APPLICATIONS OF PROMPT
ENGINEERING OBJECTIVE ASSESSMENT FINAL — NEWEST
2026/2027 ACTUAL EXAM 200 COMPLEX EXAM
QUESTIONS AND CORRECT VERIFIED ANSWERS |
ALREADY GRADED A+ (MOST RECENT!!)
1. A healthcare organization is implementing an AI system to analyze patient
records. The system performs well on the training data but fails to generalize
to new patient populations. Which fundamental limitation of AI does this
scenario best illustrate?
A) Lack of emotional intelligence
B) Dependence on training data
C) Limited computational power
D) Excessive algorithmic complexity
Answer: B) Dependence on training data
Rationale: AI systems are fundamentally dependent on their training data.
When the training data does not represent the broader population, the model
fails to generalize. This is a core limitation of AI—it cannot reason beyond the
patterns it has learned from its training data.
, 2
2. A marketing team uses an AI chatbot to handle customer inquiries. The
chatbot correctly identifies that a customer is asking about a product return
but suggests a return policy for a completely different product. This error is
best classified as:
A) True negative
B) True positive
C) False positive
D) False negative
Answer: C) False positive
Rationale: A false positive occurs when the model incorrectly generates
outputs that are not relevant or accurate to the input or task. The chatbot
correctly identified the intent (return) but provided incorrect information
(wrong product's policy), making it a false positive.
3. A data scientist is training a model to classify customer emails. She provides
500 labeled examples of spam and non-spam emails. This is an example of
which learning paradigm?
A) Unsupervised learning
B) Reinforcement learning
C) Supervised learning
D) Semi-supervised learning
Answer: C) Supervised learning
Rationale: Supervised learning uses labeled data where the model learns to
map inputs to outputs based on examples. The data scientist provided labeled
examples (spam/non-spam), which is the defining characteristic of supervised
learning.
, 3
4. A researcher wants to use an LLM to solve a complex math problem. She
asks the model to "show all your work and explain each step." This prompt
engineering technique is called:
A) Zero-shot prompting
B) Few-shot prompting
C) Chain-of-thought prompting
D) Self-consistency prompting
Answer: C) Chain-of-thought prompting
Rationale: Chain-of-thought (CoT) prompting encourages LLMs to explain their
reasoning and reveal intermediate steps. By asking the model to "show all
your work," the researcher is using CoT to improve the model's reasoning and
accuracy.
5. A company is developing an AI recruitment tool. The tool is trained on 10
years of hiring data from a male-dominated industry. Which type of bias is
most likely to affect this system?
A) Measurement bias
B) Sampling bias
C) Algorithmic bias
D) Confirmation bias
Answer: C) Algorithmic bias
Rationale: Algorithmic bias arises from assumptions in AI design that can
amplify existing data biases. The historical hiring data likely contains gender
bias, and the algorithm may learn and perpetuate these patterns, favoring
male candidates.
, 4
6. A user is interacting with an AI image generation tool. They start with "a
cat" and then refine to "a fluffy orange tabby cat sitting on a windowsill at
sunset." This process is best described as:
A) Zero-shot prompting
B) Iterative prompting
C) Chain-of-thought prompting
D) Few-shot prompting
Answer: B) Iterative prompting
Rationale: Iterative prompting involves refining prompts through successive
iterations by adding additional details, modifiers, or parameters to improve
the quality or specificity of the generated images. The user starts simple and
progressively adds detail.
7. An AI system is being used to diagnose diseases. A patient's rare condition
is missed because the training data contained very few examples of that
condition. This is an example of:
A) Algorithmic bias
B) Sampling bias
C) Measurement bias
D) Confirmation bias
