1. In the smart speaker pipeline, which step involves no machine learning at all?
2. Three chained machine learning steps are each 95% accurate. What end-to-end accuracy should be expected, and why does it matter?
3. What is the most transferable habit taught by the smart speaker case study?
4. What is the "long tail" in the self-driving example?
5. A vendor demonstrates their document system working perfectly on ten sample invoices. What should you ask?
6. Why should early projects favour closed problems over open ones?
7. In sampling a hundred real cases, a manager finds 40 straightforward, 40 awkward and 20 genuinely hard. What does this suggest?
8. Why does a realistic capability-building sequence put a working pilot before writing an AI strategy?
9. Why does the sequence emphasise training non-technical people broadly?
10. What does the AI strategy question "what data do we have that others do not?" identify?
11. Zillow Offers shut down in November 2021 with a $304 million inventory write-down and losses over $500 million, after roughly doubling home purchases twice during 2021. Which pitfall does this illustrate?
12. Which of these are warning signs that an AI initiative is drifting? Select all that apply.