In layperson’s terms, prompt engineering would mean providing efficient inputs or questions to a generative AI model to provide you with the best results you are looking for.
What is Prompt Engineering?
In this situation, think about it. There are so many different kinds of generative AI models. You can think about an image generator model. For example, where you put certain kinds of captains, and it can generate the exact kind of image output just out of thin air. You would have seen many people putting out pictures generated from dally things like a dog wearing a space suit and flying in the air, so these are some of the specific prompts you will give that model. Hence, it can generate the concrete results you’re looking for.
Why it is important
If you have ever done this in the past, When you’re using Google Search And an app like ChatGPT or any other generative AI system, how do you optimize the desired results? This is what is called prompt engineering.
How to learn prompt engineering
You will need more than one source to learn the art of prompt engineering because many components go into prompt engineering. You need to understand the capabilities of that particular system inside and out, which means what kinds of results this system can produce, what kinds of output formats it can produce, what extent of knowledge this system has, what kinds of questions or language components that it can understand, etc.
For example, if you go to ChatGPT and ask it to give me a diagram of the ML ops pipeline, it won’t give that. It will say that I’m a language model and I only produce text-based results. So you need to understand that the knowledge breadth or output breadth that ChatGPT has is restricted to text output.
Similarly, it’s said that ChatGPT is trained based on data up until 2021, which means that any questions you ask it. Which date after that, it will not be able to answer. There are restrictions, subject matter, or boundaries for each of these applications that you need to be well aware of to judge these kinds of applications best.
Example
When using ChatGPT, you can ask ChatGPT to act like a person, so you can ask ChatGPT to be your financial advisor and say, Hey, imagine you’re my financial advisor. I’m making money, and what is my monthly expenditure? I have responsibilities toward my kids, wife, family, etc. And now help me out with sound financial planning. ChatGPT understands how to get into a specific persona and gives you subject matter expertise from that particular point of view. So that’s an excellent example of prompt engineering on utilizing the best knowledge from ChatGPT.
Also Read; How Artificial Intelligence Impacts the Job Market
1 thought on “What is Prompt Engineering?”