Skip to Content
Home └Build Your Personal AI Digital Twin: The Owner's Interview └Final Quiz: Build Your Personal AI Digital Twin

Build Your Personal AI Digital Twin: The Owner's Interview

0 %

Completed

Course content

Final Quiz: Build Your Personal AI Digital Twin

10 XP
Prev Next
Fullscreen Share

Test your knowledge with this 8-question quiz.

Rating
0 0

There are no comments for now.

Join this Course
to be the first to leave a comment.

1. Your twin receives an email from a longtime client asking it to confirm a Friday delivery date. Your profile lists committing delivery dates as approval-required. What should the twin do?
Send a short confirmation itself, since the tone rules make it sound like you Reply declining the date on your behalf, to protect the commitment limit Prepare the confirmation as a draft and raise an approval request with reasoning and a deadline, sending nothing Decide using the values ranking, since client trust outranks profit
2. Why must values be ranked in order, not just listed?
When two good values collide, the twin needs a tiebreaker, or it will improvise one at the worst moment LLMs ignore unranked lists The profile format allows exactly five values Rankings make the profile shorter and cheaper to store
3. In this course, the phrase draft-not-approved refers to what rule?
The twin may send its own drafts after a short delay The twin sends anything internal without approval The twin writes and revises the message, and only the owner sends it The twin marks messages DRAFT so clients know they are unofficial
4. How are the twin's authority boundaries enforced in practice?
The system prompt tells the twin to behave well Written approval lists wired into an approval workflow, checked against an audit log Spot-checking outputs when something feels off Trust - the twin was configured with your values, so it polices itself
5. What is shadow mode?
The twin acts invisibly as a backup while the owner is on vacation Two AI models answer the same prompt and you pick the better output The twin works only after hours to save money The twin proposes recommendations while you keep acting normally, and you compare its judgment with your real decisions over an observation period
6. What failure does the never-claim list prevent?
Sending client emails without approval Forgetting to escalate true emergencies Booking meetings into protected calendar blocks Presenting confident-sounding expertise on topics the owner declared off-limits
7. Which set of items belongs in a complete approval request?
The message text plus an AI confidence score The exact action, the reason, options considered, what happens if declined, the risk in one sentence, and the deadline A full copy of the context window for the record Just the category and the word approve - details live in the profile
8. Under scope and boundaries, what is typically deferred to a later phase?
Client email triage and meeting briefs Approval requests and audit logging Technical summaries using your trusted-source list The personal domain - family matters, personal finance, health - plus anything whose rules you have not written yet