Question 1

Nina’s first step when selecting an AI model:

❌ Deploy the model
✅ Write a clear definition of the use case and requirements
❌ Select cheapest
❌ Choose highest-rated

Explanation:
A clear use case ensures the model selection aligns with actual business needs.

---

Question 2

Priya wants the model to continue performing well. What should she do?

✅ Continuously test and update the model as needed
❌ Skip monitoring
❌ Rely on initial tests
❌ Avoid switching models

Explanation:
Continuous evaluation ensures the model adapts to changing needs and data.

---

Question 3

Why compare different AI models using the same prompt?

❌ Eliminates guardrails
❌ Ensures faster integration
❌ Reduces deployment complexity
✅ Allows consistent evaluation across models

Explanation:
Using the same prompt provides a fair comparison of performance and output quality.

---

Question 4

What should Leila analyze when comparing AI models?

❌ API update frequency
✅ Performance, speed, reliability, cost
❌ Popularity
❌ Training dataset size

Explanation:
These four metrics determine real-world suitability of the model.

---

Question 5

How can Tania improve domain-specific performance?

✅ Fine-tune the model with domain-specific data
❌ Zero-shot prompting
❌ Prompt chaining
❌ Random prompting

Explanation:
Fine-tuning adapts the model to specialized terminology and tasks.

---

Question 6

Devon must keep sensitive data inside company infrastructure. Best approach?

❌ External datasets
❌ APIs
❌ Public cloud
✅ Run the AI model locally

Explanation:
Local or on-prem deployment ensures data never leaves company boundaries.

---

Question 7

Which method gives the model examples to follow?

❌ Chain-of-thought
❌ Zero-shot
❌ Random prompting
✅ Few-shot prompting

Explanation:
Few-shot prompting includes sample inputs/outputs to guide model behavior.

---

🧾 Summary Table