Introduction To Prompt Engineering.
ChatGPT has revolutionized the world. The chat based AI system can help us in many aspects.
But to get effective response from chat based AI's like ChatGPT, Bard, Bing Chat etc, it is important that we ask them proper questions, also known as 'prompt'. And, the process and practice of improving the quality of outputs based on better prompts is called as 'Prompt Engineering'.
Here's a basic introduction to get you familiarized with best practices of Prompt Engineering that would help you get better responses from ChatGPT like systems.
PROMPT RECOMMENDATIONS:
1. The more specific your prompt is, the more accurate the response is likely to be.
Example:
Less specific:
"Tell me about heart disease."
More specific:
"What are the most common risk factors for coronary artery disease?
2. Describe your specific setting and provide the context!
Example
I'm writing an article about tips and tricks for prompt engineering for people working in Financial Advisory Services. Can you please list a few of those tips and tricks with some specific prompt examples?
3. Experiment with different prompt styles!
Example
Direct Question:
"What are the warning signs of recession?"
Request for List:
"List all the potential warning signs of recession"
Request for Summary:
"Summarize the key early signs of recession"
"Explain the warning signs of recession like I'm five."
4. Identify the overall goal of your prompt first!
Example
"I'd like to get a short list of 5 ideas for a Youtube video on the future of Al in Financial services."
5. Ask it to play roles!
Example
"Act as a Data Scientist and explain Prompt Engineering to a Bank manager."
"Act as my nutritionist and give me tips about a balanced healthy diet."
6. Iterate and refine your question and ask AI chat to modify the output based on its previous response.
Example:
First prompt:
"How are you feeling today?"
Modified prompt:
"On a scale of 1-10, how would you rate your stress levels today and what specific event contributed to it?"
7. Use chat history, if available.
AI chatbots like ChatGPT maintains chat history, also called as threads.
You can easily go back to a specific discussion by clicking on the proper thread in the left column. This way, you don't have to start all over again but can just continue a discussion you already had with ChatGPT.
8. Ask Open-ended Questions as those often yield more comprehensive responses.
Example
Closed question:
"Is exercise important for patients with osteoporosis?"
Open question:
"How does regular physical activity benefit patients with osteoporosis?"
9. Request specific examples
If there is something you are not satisfied with or don't understand based on the AI chatbot's response, first of all, tell it that you don't understand the answer and ask it to provide an example.
Example
Initial prompt:
"Could you explain the common side effects of this medication?"
Refined prompt:
"I'm not clear about the side effects of this medication. Can you provide specific examples of common side effects patients have experienced?"
10. If you're asking about a process or timeline, specify that in your prompt.
Example
Without time reference:
"Describe the healing process after knee surgery."
With time reference:
"What can a patient typically expect during the first six weeks of healing after knee surgery?"
11. Set Realistic Expectations:
Example
Unrealistic Prompt:
"What's the latest research published this month about Alzheimer's?"
Realistic Prompt:
"What were some of the major research breakthroughs in Alzheimer's treatment up until 2021?"
This article introduced you to the best practices of Prompt Engineering. Following the practices will help you to get better responses from AI based chatbots like ChatGPT, Bard, Bing Chat etc..