Harnessing the power of Artificial Intelligence (AI) can revolutionize operations and enhance customer experiences in the evolving landscape of mortgage banking.
One of the critical skills for leveraging AI effectively is prompt engineering.
This blog post will explore what prompt engineering is, why it’s crucial, how iterating through feedback refines results, and provide tips and resources to help you become proficient in this essential skill.
What is Prompt Engineering?
Prompt engineering is the art and science of crafting inputs (prompts) for generative AI models to produce the desired output.
In simple terms, it involves designing questions or statements in a way that guides the AI to deliver accurate, relevant, and high-quality responses.
This skill is particularly important when using text-based AI tools like ChatGPT or Google’s Gemini, which rely heavily on well-constructed prompts to generate meaningful content.
Why is Prompt Engineering Important?
Precision and Relevance
Well-crafted prompts ensure that the AI provides precise and relevant answers, minimizing the need for extensive editing and corrections.
Efficiency
Effective prompts save time by reducing the number of iterations needed to get the desired result.
Improved Decision-Making
Accurate prompts lead to better data analysis and insights, supporting informed decision-making in mortgage banking.
Start by setting the stage
Let the tool know the specific context or role you want it to play in preparing its answers. This step helps generate responses targeting the intended audience’s needs and expectations.
Example of Setting Context
“You are an expert in crafting email marketing messages from loan officers to people they have never contacted before. Your subject lines drive up open rates and your content is easy to scan.”
Example of Setting Audience Consideration
“The user of your output will be Loan Officers and marketing people with the lender company who will be leveraging using their CRM to send the messages.”
Iterating Through Feedback: Shaping Final Results
The process of prompt engineering is iterative. Here’s how it works:
Initial Prompt
Start with a basic prompt to get the AI’s first response.
Analyze Output
Review the AI’s response for accuracy, relevance, and completeness.
Refine Prompt
Modify the prompt based on the feedback. This might involve clarifying the question, adding context, or specifying the format of the response.
You don’t have to rewrite the prompt from scratch, but instead you can respond (in the same thread) asking for revisions. Remember, its name includes the word “Chat” so treat it like you are having a conversation.
Tip: You don’t have to be polite
Most people start off treating an AI tool like a person, and this is because it is going to treat you like a person. For example, many responses from a prompt will start with “Sure, I’d love to help you with…” just like a person may say.
But, you need to remember that it is not a person, but instead a computer that is using advanced processing technics to generate responses based on mathematical probabilities. It doesn’t have emotions.
It also learns based on your direct feedback, so being straight forward with your critique is actually a good thing.
It is not uncommon for someone here at Wilqo to reply to an output with something like “that’s a terrible answer, redo it and this time organize your results in a way that is easier to understand and consume.”
You’ll be shocked at how much better the revision will be.
Repeat
Iterate this process until the AI’s output meets your expectations, but also know when to draw the line because you will likely never get the perfect output.
By continuously refining your prompts based on the AI’s feedback, you can fine-tune the results to be more aligned with your requirements. This iterative approach helps in understanding how the AI interprets different inputs and improves your ability to craft effective prompts.
Real Life Example from one of Wilqo’s Sales Reps
Initial Prompt
Initial Response
Prompt #2
Response #2
Prompt #3
Response #3
Total time spent by sales rep creating this email:
Under 2 minutes
Effectiveness:
Suzy scheduled a call (we used the “Easing the…” subject line) and everything is on track.
Tips and Tricks for Structuring Prompts for Success
Be Specific
Clearly define what you need. Vague prompts can lead to broad or irrelevant responses. For example, instead of asking, “Explain mortgage loans,” specify, “Explain the benefits of fixed-rate mortgage loans for first-time homebuyers.”
Provide Context
Give the AI enough background information. For instance, mention the target audience or the purpose of the content.
Use Constraints
Limit the length or format of the response to ensure concise and structured outputs. For example, “Summarize the key points of this contract in bullet points.”
Don’t Constrain Your Thinking
If you can think the thought, it is worth asking ChatGPT to see if it has a response for you. For example, include the instruction to “explain this like a specific youtuber” or “with the target reader being only high school educated.”
Ask Follow-Up Questions
Break down complex queries into smaller, manageable parts. Instead of asking for a full report in one go, start with, “List the main sections to include in a mortgage application report.”
Iterate and Improve
Use the AI’s initial responses to refine your prompts. This helps in gradually improving the quality of the output.
Experiment with Different Phrasings
Try various ways of asking the same question to see which one yields the best result.
Resources for Learning Prompt Engineering
To become proficient in prompt engineering, consider exploring the following resources:
OpenAI Documentation
Provides comprehensive guides and examples for using AI models like ChatGPT (https://beta.openai.com/docs/).
Google AI Resources
Offers tutorials and best practices for using Google’s AI tools (https://ai.google/).
AI Community Forums
Join forums and discussion groups where you can share tips and learn from other AI users. Websites like Reddit and Stack Overflow have active AI communities.
Online Courses
Platforms like Coursera, Udemy, and LinkedIn Learning offer courses on AI and prompt engineering.
Blogs and Articles
Follow AI-focused blogs (like this one from Wilqo) and articles to stay updated on the latest trends and techniques in prompt engineering.
Conclusion
Mastering prompt engineering is essential for mortgage bankers looking to leverage AI tools effectively. By understanding the principles of crafting precise and relevant prompts, iterating through feedback, and utilizing available resources, you can significantly enhance the outputs generated by AI. This not only improves efficiency but also supports better decision-making and innovation in your business.
Investing time in learning and refining your prompt engineering skills will pay off, enabling you to unlock the full potential of generative AI and stay ahead in the competitive mortgage banking industry.
Share this
Basic AI Tool Selection for Mortgage Bankers
How Mortgage Bankers Can Use ChatGPT to Organize Checklists
