Purpose
Use this lesson to diagnose a bad result by locating the missing instruction. The aim is to produce work that a colleague can review and act on, not merely text that sounds convincing.
Lesson Explanation
Scenario: The AI wrote a long blog post when you wanted a short LinkedIn post.
Better instruction: Rewrite the instruction: produce one LinkedIn post for operations leaders, 90–120 words, with one practical takeaway and no hashtags.
Do not endlessly regenerate the same vague prompt. Compare the output with the intended result, identify the absent instruction—audience, length, format, source, or boundary—and change that element. Keep a record of the version that works. The revised post is for operations leaders, 90–120 words, practical, and contains no hashtags.
Work in a short cycle: provide the relevant input, state the result and limits, inspect the draft against the acceptance criteria, then revise the instruction when a requirement is missing. Keep the final human decision with the person responsible for the work.
Practice Questions
-
An AI produces a five-paragraph formal report when a two-sentence casual summary was wanted. What does this lesson recommend, instead of simply regenerating with the same prompt?
View Answer
Comparing the output with the intended result, identifying the specific missing instruction (length, tone, or format), and changing that element.
-
What four types of missing elements does this lesson name as common causes of a weak prompt result?
View Answer
Audience, length, format, source, or boundary (any combination named).
-
A prompt produces a result that’s the wrong length. Based on this lesson’s debugging approach, what should be changed first?
View Answer
The specific missing element causing the mismatch – in this case, adding an explicit length constraint, rather than regenerating with the same vague prompt.
-
What does this lesson recommend keeping once you find a prompt version that works?
View Answer
A record of the version that works.
-
A colleague keeps clicking “regenerate” on the same vague prompt, hoping for a better result by chance. What does this lesson say about this approach?
View Answer
It advises against endlessly regenerating the same vague prompt, and instead recommends diagnosing and fixing the specific missing element.
-
An AI-generated email is too formal for its intended casual audience. Applying this lesson’s debugging process, what is the first diagnostic step?
View Answer
Comparing the output with the intended result to identify what specifically caused the mismatch – in this case, likely a missing tone/audience specification.
-
A prompt produces content that includes information from an unintended, irrelevant source. What category of missing element, per this lesson, does this likely reflect?
View Answer
Source – the prompt likely didn’t specify which source material to use.
-
Why does this lesson recommend identifying “the absent instruction” rather than rewriting the entire prompt from scratch each time?
View Answer
Precisely diagnosing the specific gap allows a targeted fix, which is more efficient and reliable than repeatedly regenerating with the same vague, unrevised prompt.
-
A request produces an overly long response despite multiple attempts with the same prompt. What specific missing element, per this lesson’s categories, is most likely the cause?
View Answer
Length constraint (a missing boundary).
-
A prompt is revised to add a specific audience and format after an initial weak result. The next attempt succeeds. What principle from this lesson does this process illustrate?
View Answer
Debugging by comparing the output to the intended result, identifying the missing element, and revising that specific part of the prompt.
-
Why might keeping a record of a working prompt version matter for future tasks, according to this lesson?
View Answer
It preserves a proven, working instruction that can be reused or adapted, rather than starting the debugging process from scratch each time a similar task arises.
-
A prompt for a technical explanation is far too jargon-heavy for its intended audience, despite several regeneration attempts with the same wording. What should change, based on this lesson’s debugging approach?
View Answer
The prompt itself should be revised to specify audience and reading level explicitly, rather than regenerating the same unrevised instruction repeatedly.