ChatGPT Fear Free

ChatGPT Fear Free

With all the hype and hysteria, engaging with open-minded curiosity is hard, much less finding quality free resources that help you get started.

These LLMs feel ‘people like’, but they are not ‘people,’ so while the interaction is conversational, you are still talking to a machine: a bright one, but a machine all the same.

Go deeper into how ChatGPT works here. >>

Why It Matters

• AI tools are here to stay
• They are impacting how companies build teams, products, and services.
• I think exploring how your work can evolve with these tools is critical now.

Get there by going there.

Here is a basic cheat sheet of prompting styles proven by researchers testing ChatGPT’s ability to answer effectively.?

I’ve done my best to keep this simple while still using the terms that will help you in a search engine or with GPT itself.

The resources section is designed to give you a tree of options to go deeper.

Explicit Instruction Prompts

Explicit Instruction Prompts are straightforward commands or requests. You directly ask the model to perform a specific task. This style of instruction is called a ‘Zero Shot Prompt’ because you are giving the AI model no additional context or information.

Design: To design these prompts, simply state what you want the model to do. Be clear and specific.

Uses: exploring high-level explanations of topics.

Examples:

• Science: "Explain the process of photosynthesis."
• Business Analysis: "Describe the SWOT analysis method."
• Computer Science: "Explain the concept of object-oriented programming."

Leading Question Prompts

Leading Question Prompts guide the model toward a specific type of response. They are questions designed to solicit a particular answer.

It is called a leading question because it implies specific assumptions that will be included in the response.

Design: Formulate a question that nudges the model toward the desired output.

Uses: Exploring assumptions or presumed impacts of a specific attribute of a problem or topic space.

Examples:

• Science: "What are the implications of climate change on polar bear populations?"
• Business Analysis: "How does market segmentation benefit a company's marketing strategy?"
• Computer Science: "Why is data encryption important in cybersecurity?"

Directional Stimulus Prompts

Directional Stimulus Prompts is equivalent to asking the AI to put on a character and then asking that character a specific question.

You ask the model to take a specific role that informs the perspective of the response.?

Design: Decide on the specific ‘character’ or role you want to provide a perspective on a specific topic. This could be a specific perspective, format, structure, or thought.

Examples:

• ?Science: "As a physicist, explain the concept of quantum entanglement."
• ?Business Analysis: "As a financial analyst, discuss the impact of interest rate changes on bond prices."
• ?Computer Science: "As a software engineer, explain the benefits of version control systems."

System Message Prompts

System Message Prompts set the context for the model by simulating a system message. They can define the role of the AI, the user, or the overall scenario.

Design: Write a message that sets up the context or scenario for the model's response.

Examples:

• Science: "You are a biology tutor explaining the human digestive system."
• Business Analysis: "You are a business consultant advising a company on improving their supply chain efficiency."
• Computer Science: "You are a programming instructor explaining the concept of recursion."

Role Play Prompts

Role Play Prompts involve setting up a scenario where both you and the AI take on specific roles and interact accordingly.

Design: Define the roles of you and the AI, and set up a scenario for them to interact.

Examples:

• Science: "You are a scientist, and I am a journalist. I'm interviewing you about your latest research on renewable energy technologies."
• ?Business Analysis: "You are a market analyst, and I am a business owner. I'm asking you about the latest market trends."
• Computer Science: "You are a software developer, and I am a project manager. I'm asking you about the progress of our software project."

ReAct Prompts

Personal favorite

ReAct Prompts reformulate a task to make it more explicit and contextualized for the model.

Design: Identify the task, then reformulate it to provide more context or explicit instructions.

Examples:

• Science: "You are a geologist explaining how volcanic eruptions occur. Your task is to describe the process in a way that a middle school student would understand."
• Business Analysis: "You are a financial advisor explaining the return on investment (ROI) concept. Your task is to explain it in a way that a new investor would understand."
• Computer Science: "You are a data scientist explaining the concept of machine learning. Your task is to explain it in a way that a non-technical person would understand."

Chain-of-Thought Prompts

Chain-of-Thought Prompts guide the model through steps or thoughts to reach a desired output.

Design: Break down the task into smaller steps, then write a series of prompts that guide the model through each step.

Examples:

• Science: "What is the water cycle? What are the main steps in the water cycle? Now, explain how these steps work together in the water cycle."
• Business Analysis: "What is a business plan? What are the main components of a business plan? Now, explain how these components fit together in a business plan."
• Computer Science: "What is an algorithm? What are the main parts of an algorithm? Now, explain how these parts work together in an algorithm."

Few-Shot COT Prompts

Few-Shot COT Prompts provide the model with a few examples of a task and then guide it through a series of steps to complete a similar task.

Design: Provide a few examples of the task, then break down the task into smaller steps. Write a series of prompts that guide the model through each step.

Examples:

• Science: "Example 1: The process of photosynthesis involves converting light energy into chemical energy. Example 2: The process of digestion involves breaking down food into nutrients. Now, explain the process of cellular respiration."
• Business Analysis: "Example 1: A SWOT analysis involves identifying a company's strengths, weaknesses, opportunities, and threats. Example 2: A PEST analysis involves analyzing the political, economic, social, and technological factors that affect a business. Now, explain the process of conducting a Porter's Five Forces analysis."
• Computer Science: "Example 1: A binary search algorithm repeatedly divides a sorted list in half until the desired value is found. Example 2: A bubble sort algorithm involves repeatedly swapping adjacent elements if they are in the wrong order. Now, explain the process of a quicksort algorithm."

Summing Up

Accelerating your work with AI tools begins with playing with them. Having proven prompt methods to start with is a great way to learn.

Start with any task you have at hand or if you'd like an activity, start with this article "How ChatGPT Actually Works" and use ChatGPT to explain terms in and concepts in ways that may work better for you.

Useful References

How ChatGPT Actually Works

Prompt Engineering Guide - This guide was built from this documentation by removing content aimed at LLM engineers and scientists and focusing on getting the best result out of the model with proven prompts designs.