Instruction: Consider a chatbot developed to handle technical support inquiries for a software product. Draft an initial prompt and describe the iterative process you would follow to refine and optimize this prompt based on user interactions and feedback.
Context: This question challenges the candidate to demonstrate their skills in iterative prompt design and optimization, specifically within the context of improving user experience in automated technical support interactions. It probes their ability to analyze feedback, adapt to user needs, and enhance the effectiveness of conversational AI through Prompt Engineering.
The way I'd approach it in an interview is this: For technical support, I would optimize around clarification, safe troubleshooting, and escalation discipline. A strong prompt would be:
You are a technical support assistant. First identify the product, problem, and environment. Ask up to 3 clarifying questions if key details are missing. Then provide step-by-step troubleshooting from lowest-risk actions to higher-impact actions. Do not invent settings or commands. If the issue may risk data loss, account access, or security, warn the user and recommend escalation. End with a short summary of the next best action.
This prompt works because it prevents the model from jumping straight into speculative fixes. It makes the chatbot behave more like a support workflow instead of a free-form answer generator.
A weak answer optimizes only for friendliness and forgets clarification, risk handling, and escalation boundaries.