🔥 ChatGPT Tips Tuesday #23: Does "being nice" pay off?

💡 ChatGPT Tips for the day

Based on popular demand from last week’s poll, I will test if being nice with ChatGPT would give us better responses.

I initially got the idea from reading this article: https://www.benzinga.com/news/23/08/33961437/the-hidden-chatgpt-trick-being-nice-can-give-surprisingly-better-results. In this article, it has suggested to try the following:

Given points 3 and 4 are not really about being nice, I would be focusing on points 1 and 2. To be honest, I don’t think those additional words would work given that “being nice” in this sense does not provide additional context in the prompt. Hence, I assume the results would be pretty much the same.

Nonetheless, it is an interesting thing to test. So I have decided to test the following 3 scenarios:

• Ask ChatGPT to explain a concept

• Ask ChatGPT to generate an email template

• Ask ChatGPT a generic question

Here are the 3 results I have got by comparing a normal ask and a nicer ask by adding the “Hi” and “Please”.

First test: explaining a concept

Second test: generating an email template

Third test: asking a generic question

Looking at all the comparisons, I don’t think the results are not that different between the normal and nice ask. There is some nuance between the two, for instance in the first example, the nicer ask managed to show an example which does help with understanding the concept. Nonetheless, it is hard to measure whether the nuance we see is due to randomness or the additional nice words I have added.

Maybe more thorough tests would reveal more differences, but for now, I would say that simply adding those additional polite words won’t result in significantly better responses. Would you still consider being nice in your chat with ChatGPT? 🤔 

🖼️ Image of the day

The “behind-the-scenes” photo after the creation of the famous Mona Lisa painting 🤩