To understand what is prompt engineering and why we need it we will first review some important topics from the previous section:

1. A prompt or query is the input text provided to the model to initiate a conversation or request a response. It serves as the user's message or question that prompts the model to generate a text-based response;
2. The way ChatGPT responds can be influenced by how a request is formulated; if you rephrase the same question using different words, the model may generate varying responses.

Based on these two topics we can understand what prompt engineering is: it is the process of carefully crafting and structuring input prompts or queries to elicit desired responses from the model.

Let's look at the examples:

Bad prompt

This is considered "bad" for a few reasons:

1. Lack of Specificity: The prompt is very broad and doesn't specify which scientists or areas of science the user is interested in. It doesn't provide clear guidance to the model on the scope of the response;
2. Vague Expectations: It doesn't indicate what kind of information the user is seeking about scientists. Are they interested in biographies, scientific contributions, historical context, or something else? Without this clarity, the response may not meet the user's expectations;
3. Open-Ended: The prompt is open-ended, which can lead to a generic or general response that may not be particularly informative or engaging.

Let's modify this prompt to get more specific result.

Good prompt

The "good" prompt is effective because it provides specificity, context, and clear guidance. It specifies the topic (influential scientists in the field of physics) and the type of information desired (key discoveries), making it easier for ChatGPT to generate a relevant and informative response that aligns with the user's expectations.