A professor uploads their syllabus and prompts Gemini: “Generate a 10-question quiz based on this syllabus to help students understand the course’s main learning objectives and major assignment policies.” The AI generates a quiz that instead focuses on trivial, non-academic details (e.g., “What is the professor’s office number?”, “What font is the syllabus written in?”). This output should be analyzed as: (Select all that apply)
The correct choices are a failure to understand the semantic context and purpose of a syllabus and pedagogically inappropriate, as it misaligns with assessment goals.
When an artificial intelligence tool extracts trivial details like font types or office numbers instead of focusing on core learning objectives, it demonstrates a clear breakdown in semantic understanding. The model successfully read the uploaded text but failed to grasp the relative importance of the information. It missed the primary intent of your prompt, treating all text within the document as equally valuable data points rather than recognizing what functionally matters in an academic setting.
Consequently, the resulting output is pedagogically inappropriate. A well-designed assessment must align with specific learning goals. Quizzing students on administrative minutiae completely fails to measure their comprehension of the course policies or academic expectations, rendering the quiz useless for its intended educational purpose.
The other options do not accurately describe this technical failure:
- A hallucination refers to the AI inventing fake information. In this case, the office number and font style likely exist within the document; the system just prioritized the wrong facts.
- A bias against the prompter implies a systemic prejudice or unfair weighting based on demographics or specific viewpoints. Generating a poorly focused quiz is simply a failure of contextual logic, not an act of bias.
To prevent this issue, educators often need to use stricter prompt constraints, explicitly telling the model to ignore administrative headers and focus exclusively on the sections detailing grades and learning outcomes.