1. Question 1

Current limitations of AI (Select ALL):

• ❌ There are no limitations

• ✅ AI can be biased

• ✅ Explainability is hard

• ✅ AI is susceptible to adversarial attacks

• ✅ AI can display discriminatory behavior

---

2. Question 2

Goldilocks Rule of AI:

• ❌ Fear extinction

• ❌ Expect utopia

• ✅ Don’t be too optimistic or too pessimistic about AI

• ❌ AI will solve all problems

---

3. Question 3

Explainability in medical AI:

• ❌ Achieved through chatbots

• ❌ AI cannot explain anything

• ✅ Lack of explainability reduces user trust

• ❌ AI is always explainable

---

4. Question 4

If ML learns from biased text, it behaves biased:

• ✅ True

• ❌ False

---

5. Question 5

If ML learns from completely unbiased text, will it have minimal bias?

• ❌ True

• ✅ False

Explanation:
Even “neutral” data can reflect hidden societal bias, and models can generate new unintended biases.

---

6. Question 6

Recommended bias-mitigation practices (Select ALL):

• ❌ Zeroing out bias (ineffective)

• ✅ Use more inclusive / less biased data

• ❌ Adversarial attacks to change outputs

• ✅ Systematic auditing for bias

---

7. Question 7

Examples of adversarial attacks (Select ALL):

• ✅ Subtly altering an image to fool recognition

• ✅ Adding a sticker to a stop sign

• ✅ Modifying audio to trick speech recognition

• ❌ Synthesizing a fake video (that’s deepfake generation, not an adversarial attack)

---

8. Question 8

If a developing economy has a strong coffee industry, it has an advantage applying AI to it:

• ✅ True

• ❌ False

Explanation:
Strong domain expertise + data availability helps AI adoption.

---

9. Question 9

Jobs most likely displaced by AI:

• ❌ All jobs

• ❌ Non-routine work

• ✅ Routine, repetitive jobs

• ❌ Most white-collar jobs (some but not most)

---

10. Question 10

Congratulations statement:

• ✅ True 🎉

• ❌ False

---

🧾 Summary Table

---