Week 4 : Prompts

July 07, 2024

Dear DataDrivenAEC Subscribers,

In our fourth week of data literacy for architects, we explored several tricks for getting better results from chatbots like ChatGPT. Talking to a chatbot is about getting the result you want, and here are several principles that can help you communicate more clearly with the chatbot. Let’s recap!

Day 16 of 100: Ask for the source!

Ask for the source and get a formated citation can help the AI get more accurate result.

Day 17 of 100: Specify the steps first.

Specifying the steps or ask the AI to think of the steps first can help AI reason better. Step by step!

Day 18 of 100: Use Examples.

Giving an example can help the AI understand what you want a lot better. One or a few would work!

Day 19 of 100: Role Playing.

Ask the AI to adopt a specific role. Imagine you have an expert to help you with all sorts of tasks.

Day 20 of 100: More Detail.

Simply make your requirement more specific with clear instructions. It takes a bit of thinking, but then you get a better result. Iterate until you get the desired outcome.

Here is the source from OpenAI on strategies for prompt engineering:

Next week, we will talk about more resources for learning data analysis. Stay tuned for more updates as we continue to build our data literacy skills together.

You can find all the previous articles here.