1. Question 1

Type of AI used today in spam filters, translation, speech recognition:

• ✅ Artificial Narrow Intelligence (ANI)

• ❌ Artificial General Intelligence (AGI)

---

2. Question 2

AI technology for input→output mapping:

• ❌ AGI

• ❌ Reinforcement learning

• ❌ Unsupervised learning

• ✅ Supervised learning

---

3. Question 3

To build a strong supervised deep-learning speech recognition system, you need:

• ✅ A large dataset

• ❌ A small dataset

• ✅ A large neural network

• ❌ A small neural network

---

4. Question 4

Only manual labeling can create supervised data:

• ❌ True

• ✅ False

Explanation:
Labels can come from logs, existing systems, weak labels, user behavior, etc.

---

5. Question 5

Data acquisition reality:

• ❌ AI teams can produce data themselves

• ❌ Only structured data is valuable

• ✅ Some types of data are more valuable; AI team can guide acquisition

• ❌ More data always better (not always true)

---

6. Question 6

Examples of unstructured data:

• ❌ Max speed (structured)

• ✅ Pictures of scooters

• ❌ Weekly sales numbers (structured)

• ✅ Audio recordings of scooters

---

7. Question 7

Good results for a data-science project for a cat-food website:

• ❌ Neural network mimicking cat brain

• ✅ Pricing strategy slide deck

• ❌ Image dataset (not a usable business result)

• ✅ Marketing insights by breed

---

8. Question 8

Correct AI terminology:

• ✅ Deep learning is a type of machine learning

• ❌ AI is a type of deep learning

• ❌ Machine learning = Data science

• ✅ Deep learning ≈ neural networks (used interchangeably)

---

9. Question 9

What do AI companies do well?

• ❌ Strategic data acquisition alone

• ❌ Invest in data warehouses alone

• ❌ Spot automation alone

• ✅ All of the above

---

10. Question 10

If humans can do an A→B task in <1 sec, supervised learning can likely learn it:

• ✅ True

• ❌ False

---

🧾 Summary Table

---