AI-assisted practical guide. Examples are hypothetical; these are proposed editorial methods, not reported research results.
Creating a practice set that rewards students for identifying missing information prevents them from guessing or hallucinating answers. When learners are trained to recognize when a text does not contain the necessary evidence, they develop stronger critical thinking skills and a more disciplined approach to data analysis. This method shifts the goal from simply finding an answer to verifying if an answer is actually supported by the provided source.
Designing the Unanswerable Question
To implement this, you must first select a source text and identify a plausible but absent detail. The question should look identical in structure to the answerable ones to avoid tipping off the student. For example, if the text describes a company's quarterly revenue and growth, ask about the specific marketing budget used to achieve those results if that figure was never mentioned. The key is to ensure the answer is not implied or easily inferred from general knowledge. You should explicitly instruct the learner that some questions may be unanswerable based solely on the provided text and that stating this is the correct response.
Hypothetical example
Imagine a practice set based on a fictional company profile for SolarStream. The text states that SolarStream was founded in 2012 in Arizona and currently employs five hundred people. A standard question might ask for the founding year. The unanswerable question would be: Who is the current Chief Executive Officer of SolarStream? Because the text mentions the size and location but omits the leadership names, the correct student response is: The provided text does not mention the name of the CEO.
Verifying the Practice Set
Once the set is complete, you must verify that the unanswerable question is truly impossible to solve using only the source. Read the text through the lens of a student who might try to over-analyze or bring in outside knowledge. If there is any sentence that could be interpreted as a hint toward the answer, rewrite the question to be more specific. The final check involves reviewing the student's actual deliverable to ensure they did not invent a plausible answer based on assumptions. If a student provides a logical guess instead of noting the missing evidence, the prompt for the unanswerable question may need to be more clearly decoupled from any related keywords in the text.
To automate this, use a prompt such as: Create a three-question reading comprehension set based on the following text, ensuring one question cannot be answered using the text provided.
Input: A short biography of a fictional painter named Elara who lived in France and used oils.
Output: 1. Where did Elara live? 2. What medium did she use? 3. Which gallery first exhibited her work? (Unanswerable).
A human reviewer should check if the AI accidentally included the answer to the third question within the biography.