My blog, ‘Framework of Smart Prompts in 7 steps’, is a detailed narration of the information given in the following table:-
| Step | Description | Details |
|---|---|---|
| Introductory video to Smart Prompts | ||
| 1. Define the Objective | Clearly understand the goal or purpose of your prompt. | Is the prompt a text generation prompt, a question-answering prompt, a brainstorming prompt, etc. |
| 2. Identify the Audience | Make the prompt fit in with the recipient's knowledge level. | In technical matters, use jargons; for non-technical user, simple and simple enough. |
| 3. Clear Instruction | Set your prompt clear as possible, avoiding any ambiguity. | Replace "Tell me about AI" with "Explain how AI in healthcare adds value in 3 bullet points." |
| 4. Add Context | Provide some background to be able to guide the AI system in getting the best possible replies. | This can be found within the use cases themselves, prior knowledge, or examples that guide the AI to respond in certain ways. |
| 5. Set Constraints | Limit the scope or length of the AI response for better control. | For example: "Summarize no more than 100 words"; or "Identify 3 key advantages of 5G technology. |
| 6. Inspire Creativity | If needed, let them respond to open-ended or creative answers. | Example: "Imagine a future where space tourism is common. Describe what a week long trip would be like."END |
| 7. Refine and Iterate | Review, test, and improve your prompt based on initial results. | Revise wording, clarity, and limitations in accordance with what the AI shoots back in test runs. |
I am going to share with you the process of Framework of Smart Prompts in 7 steps which we have learned from the best prompt engineers. It is good to experiment these steps and you will get amazing results. You need to be good at prompt engineering.
Prompt engineers are the one who coax AI to produce better results. Their job is to write prompts and interact with the LLM(Large Language model) of the ChatGPT. The Ai companies have specific type of people in their job roles. Example:- There are backend Developers, Front end Developers, Database Engineers. We can have Prompt Engineers.
What is AI?
- Teaching computers to do, what we humans can do and beyond that. example: You want to write a You tube Script, and think of different ideas. Our brain is creative. We can come with creative content and ideas and come up with a lot of things. Training the machine for AI to do the same thing.
- It is helpful for experienced digital marketers and experienced business people, working professionals.
What is Prompt?
- It is a simple text that we give to the Chatbot and it responds.
- GPT- Generative Pretrained Transformer
What is Prompt Engineering?
How to engineer the best Prompts, the best output from AI. Learning how to talk to AI to get it to do what you want. There is a concept of LLM(Large Language Model), Chat GPT is trained on a set of 100 trillion Data Sets.
People are using Chat GPT and asking number of questions to AI, it is getting better and better.
One day AI will be more smarter than the human brain.
How to write Better Prompts in Chat GPT(Framework of Smart Prompts in 7 steps)
You can Apply the following 7 steps:
- The context of the Prompt is to be set.
Set the scene. Think about the context in which it is used before diving into the prompt. Setting a specific field or topic for the model to focus on, will help it to understand the purpose of the conversation and help more relevant answers.
- Define a Clear Task
Define a task. Once the context is established give the model a clear task to complete.
For Example: If the prompt is about career advice, the task could be ‘provide the best advice for changing careers’.
- Being Specific is important
Include important details such as the goal, the starting and the ending points, characters involved in any relevant background information should be specified. If the prompt is too vague, it could result in off topic, irrelevant or consistent responses.
Example: You have to create a YouTube video of Weight Loss.
Generate a YouTube Video script for a person of 100 kgs. Being over weight and it is important for him to get into the BMI(Body Mass Index) category. What he has to do is to get into the BMI category. We are giving a specific target to it. One is the starting point and other is the ending point. So that Chat GPT will give you that script. Example: Overweight men or overweight Women or overweight Kids. People who are working in the IT industry who are in the age of 25 to 40.
- Asking Questions is important
To ensure the output is relevant and helpful. Ask specific questions within the prompt. This will the model a better understanding of what you are looking for. The Chat GPT will ask questions that what is the target audience.
- Output is to be considered
Take a look at the output. Be niche specific or have a deeper output. If it is not what you were expecting that lacks details then its time to refine the prompt. Keep Improving the prompt.
- Refine and Repeat
Each repetition differs an opportunity to return or improve the prompt. For example: If the Chat GPT generates an off topic response, you could add more specific instructions or additional content to the prompt.
If you get a good output and get what you are expecting, it is fantastic. You can use it for whatever purpose you want, like Facebook Ad or Youtube script.
You can ask and and go deeper into the prompt again. You can keep asking the better questions and keep adding questions. Once you do that, you will get better outputs.
- Reuse refined prompts
Reuse the refined prompt for similar context use cases. Once you get the perfect output, then you can reuse the same kind of prompts for the next Use Cases. Example: you could be a content creator or feedback advertiser. You can have a different context. May be you have a client who is into health category or business category.
Once you understand that it is a perfect kind of a prompt that has come, what you have to do is to reuse it, change its context, change industry, Target audience, just change 5 to 6 words. This is reusing the same prompts. you will get the best outputs.
You will require time to practice these above steps. It is not complicated. You have to practice 7 to 10 prompts. It will come naturally to you.
