最終試験

合格には % 以上のスコアが必要です。

問題 /

What is the single biggest lever for getting better AI output?

Using a more expensive modelSetting a clear role for the AI (e.g. 'You are a senior security reviewer…')Asking shorter questionsAvoiding any context

問題 /

Why add explicit constraints to a prompt?

To make the model slowerTo force the AI to make tradeoff decisions inside the box you definedTo use more tokensConstraints don't help

問題 /

Which is the most useful kind of context to provide?

ComplimentsConcrete examples of inputs and the desired outputsA long company historyA list of unrelated competitors

問題 /

When the first answer is almost-but-not-quite right, what's the negotiator move?

Start a brand new chatIterate: keep the good parts, point at what's wrong, ask for a targeted revisionSwitch to a different modelAccept the answer as-is

問題 /

What is the difference between a 'before' and 'after' AI prompt?

The 'after' prompt is shorterThe 'after' prompt sets a role, gives constraints, provides context, and shows examplesThe 'after' prompt uses more emojisThere is no difference

問題 /

Why is your AI cheat sheet a personal artifact, not a generic template?

Because copyrightBecause your work, voice, and standards are unique — your cheat sheet encodes themBecause templates don't existBecause the AI prefers it