Lesson 5 Quiz — GenAI Capabilities, Limitations & Business Value (10 easy + 10 hard)
20 questions.
Q1easymcq
A model confidently states a fact that is completely false. What is this called?
Q2easymcq
What does it mean that GenAI output is 'nondeterministic'?
Q3easymcq
Which of these is a genuine advantage of GenAI over a hard-coded rules engine?
Q4easymcq
Which is a 'business value' metric rather than a technical model metric?
Q5easymcq
ARPU stands for:
Q6easymcq
Why is it hard to explain exactly why an FM produced a specific output?
Q7easymcq
Raising a model's temperature setting during testing mainly demonstrates which limitation?
Q8easymcq
Which model selection factor concerns where a model's data is stored/processed to meet legal requirements?
Q9easymcq
A customer support chatbot keeps customers engaged and answers correctly, leading customers to stay subscribed longer. Which business metric captures this long-run effect?
Q10easymcq
True or false: a model can have excellent precision and recall and still be a business failure.
Q11hardmcq
A legal-research assistant confidently cites a court case that does not exist. Which limitation is this, specifically?
Q12hardmulti
Which of the following are genuine GenAI limitations covered in this lesson?
Select all that apply.
Q13hardmcq
A team picks the cheapest available FM for a low-latency, high-throughput fraud-screening feature, ignoring its poor reasoning performance. What tradeoff did they mishandle?
Q14hardordering
Order these from most 'business value oriented' to most 'technical model oriented'.
1.Precision
2.F1 Score
3.Return on Investment (ROI)
4.Conversion rate
Q15hardmatching
Match each business value metric to what it measures.
ROI
Conversion rate
ARPU
Customer Lifetime Value
Q16hardmcq
A healthcare chatbot must justify every recommendation to a compliance auditor. Which GenAI limitation is the biggest obstacle here?
Q17hardmcq
A GenAI support bot has excellent accuracy but its per-conversation cost exceeds the value of resolved tickets. What should the business conclude?
Q18hardmulti
Which of these are valid model selection factors when choosing an FM for a use case?
Select all that apply.
Q19hardmcq
Two runs of the same prompt against the same model, same settings, produce slightly different wording. Is this necessarily a bug?
Q20hardmcq
A team wants to reduce hallucination risk for a document-Q&A assistant without retraining the model. What is the most directly relevant technique (previewed here, detailed in Lesson 7)?