Part One: Smart Prompting using a free LLM

1. Title

From Vague Ideas to Valuable Insights: A Guide to Smart Prompting

2. Objective

The objective of this assignment is to provide students with a hands-on experience in structured and systematic prompting. Students will learn how to iteratively refine a simple request into a sophisticated prompt, guiding a freely available LLM to produce a high-quality, actionable business analysis.

3. Tools/Resources Needed

• Web Browser
• Access to at least two freely available LLMs (e.g., ChatGPT, Google Gemini, Anthropic's Claude).
• A text editor or word processor (e.g., Google Docs, Microsoft Word) for drafting prompts and compiling the final report.

4. Description of Assignment

Prompting is the art and science of communicating with an LLM to get the desired output. A simple question gets a simple answer; a well-crafted prompt can produce nuanced, structured, and insightful results. In this lab, you will act as a junior business strategist for a fictional company and use an LLM as your research assistant. Your goal is to create a mini-report on a strategic challenge, but you will achieve this by progressively improving your prompts.

This exercise will teach you several key concepts:

• Systematic Prompting: Adding roles, context, and format requirements.
• Chain-of-Thought (CoT) Prompting: Asking the model to "think step-by-step" to improve its reasoning.
• ReAct Prompting: Structuring a prompt to make the model show its work (Reasoning + Action).
• Contextual Information & Context Windows: Understanding how much information a model can remember and how to work around its limits.
• Problem Tracing: Identifying why a prompt failed and how to fix it.

Note on Free LLMs and Context Windows: Freely available LLMs have limits on how much text they can process in a single conversation (the "context window"). If you run into a limit or the model starts "forgetting" earlier parts of your conversation, simply start a new chat session with your most refined prompt. If issues persist, try switching to a different LLM.

Tasks