What is the primary goal of the prompt engineering exercise described?	To improve an existing prompt to extract all topics from a scholarly article passage into a JSON array of strings.
What was the initial average score of the poor starter prompt?	2.8
What technique was used to immediately improve the prompt's adherence to the JSON output format?	Being simple and direct in the prompt.
How should a prompt be phrased when expecting a specific structured output like a JSON array?	Be simple and direct, explicitly stating the desired format (e.g., "Extract key topics... into a JSON array of strings").
What structural technique was added to the prompt to clearly delineate the input text?	Using XML tags (e.g., `<text>...</text>`) around the content property.
Why are XML tags beneficial when structuring a prompt?	They create a clearer connection between the instruction and the interpolated input data.
What is the benefit of structuring the prompt using a series of steps?	It makes the model more specific about the required actions and output process.
List the steps suggested for the prompt to ensure correct topic extraction.	1. Closely examine the provided text. 2. Identify each topic mentioned. 3. Add each topic to a JSON array. 4. Respond with the JSON array.
What advanced prompting techniques were mentioned as optional additions to the prompt?	One-shot or multi-shot prompting.
What file is generated when running the evaluation, and what is its purpose?	output.html; it is used to understand the reasoning behind the output's poor grading.