Phase 4 · Gemini · Level 3 · Power User
Gemini: Power User Quiz · 10 min
Why it matters A quick check of your judgement before you move on. There's no penalty for a low score, and you can retake it, or take it first to test out of a level you already know.
These aren't trivia. They're the judgement calls a Power User faces: designing Gems a whole team can trust, using NotebookLM's source-locked answers, and staying critical about Gemini in Chrome, scheduled actions and generated media. Pick the best option for each scenario, and you'll see an explanation after each one.
Passing this checkpoint: work through the level first, then score 70% or more here.
Skipping this level: already confident? Take this cold and score 80% or more to test out and jump ahead. Below that, nothing is lost, you just study the level as normal.
1. You're writing instructions for a Gem the whole ops team will share. Which single line does most to make it safe for colleagues to use unsupervised? A line making the tone warmer and more friendly. A line telling it what to do when the answer isn't in its files: don't guess, say what's missing, and name the file or detail needed. A line listing every product Fernway sells. A line asking it to always answer as fully as possible. 2. You have a fixed set of documents (a project brief, notes and a policy) and you need answers you can trace back to the exact passage to quote in a meeting. Which tool fits best? An ordinary Gemini chat, because it knows more than your documents. A NotebookLM notebook, because it answers only from your sources and links each claim to the passage it came from. Deep Research, because it browses the open web thoroughly. Image generation, to make the meeting slides. 3. You generate a NotebookLM Audio Overview of a document set and it clearly states a specific budget figure. You want to put that figure in a proposal. What's the wise move? Use it, the Audio Overview is generated from your own sources, so it must be accurate. Open the notebook, find the grounded answer and its citation, and confirm the figure against the actual source before using it. Ask the Audio Overview to repeat the figure and trust it the second time. Discard the whole notebook because the audio can't be trusted. 4. Gemini in Chrome can summarise the page you're on. You're viewing a page that shows a customer's account details. Is it fine to ask it to summarise? Yes, it's just reading a page that's already open, so nothing is shared. No, to summarise the page it works from that page's content, so pointing it at confidential data is like pasting that data into a chat. Yes, as long as you delete the summary afterwards. No, because Gemini in Chrome can never read any page. 5. You set a scheduled action that emails you a daily digest of your calendar and unread mail. After a fortnight of it being right, you glance at it each morning and act on it without checking. What's the risk? None, two weeks of accuracy proves it's reliable from now on. A misread email or a slipped-in error will arrive with the same calm confidence as the accurate days, and you're now primed to act on it unchecked. Scheduled actions stop working after two weeks anyway. The digest will start sending itself to other people. 6. You ask Gemini to generate an infographic showing Fernway's regional sales, and it produces a polished chart with specific percentages. How should you treat those numbers? Use them, the chart looks professional and came from Gemini, so the figures are calculated correctly. Treat the figures as invented until proven otherwise: image generation makes up plausible-looking numbers and labels, so build anything factual from your real data yourself. Trust the numbers but redraw the chart to look nicer. Assume the percentages are right if they add up to 100. 7. A colleague wants to use a Gemini-generated image of a 'happy customer' as a real photo on Fernway's public website. What's the right response? Go ahead, if it looks realistic, no one will know it's generated. Presenting a generated image as a genuine photograph is a trust and compliance problem, and commercial use raises rights questions. This needs disclosure and a check with whoever owns brand and legal. It's fine as long as you don't tell anyone it's AI. Generated images can never be used anywhere, so bin the idea entirely. 8. You're curating the knowledge files for a shared Gem. You have this quarter's live figures, last quarter's figures, three overlapping drafts of a policy, and the current policy. What should you attach? All of them, more files means the Gem knows more and answers better. Only the current, necessary files, because a stale policy or old figures in the set produce confident wrong answers. None, a Gem shouldn't use knowledge files at all. Only the oldest versions, since they've been around longest. 9. You've built both a file-anchored Gem and a NotebookLM notebook over the same Fernway paperwork. You now need to draft the same style of customer update every week, in Fernway's voice. Which do you reach for, and why? The notebook, because citations make every draft more trustworthy. The Gem, because it bakes in the role, voice and format for a recurring, on-brief drafting job done in one click. Neither, recurring drafts should always be written from scratch. Whichever you happened to open first. 10. Before letting the team use a new shared Gem, what's the most valuable thing to do? Ask it a few questions you designed it to handle well, and if they work, roll it out. Deliberately try to break it (vague inputs, questions the files can't answer, requests that tempt it to invent a figure) then tighten instructions and re-test. Rebuild it two or three more times to make it production-grade. Roll it out immediately and fix problems as colleagues report them. Answered 0 of 10.
Answer every question to see your score Take the one-page cheat sheet with you Everything from this phase, distilled to a single page you can print or keep open while you work.
Open the Gemini cheat sheet