Fuwakami's Archive of Knowledge

This page is read only. You can view the source, but not change it. Ask your administrator if you think this is wrong.
====== ChatGPT ======
ChatGPT is a Large Language Model AI. That is to say that it is an AI that was trained using a large database on content available on the internet. It create replies by predicting the next token.

Some prompts are very useful to improve the output of ChatGPT:

===== Complexity and Burstiness =====
When an AI output text, it tends to write sentences of a more uniform style. That is to say that sentences length and complexity tend to be more uniform throughout a text compared to human output. To input more natural outputs, it may be useful to increase the complexity of burstiness. An example of this can be achieved using this prompt:

> When it comes to writing content, two factors are crucial, "perplexity" and "burstiness."Perplexity measures the complexity of text. Separately, burstiness compares the variations of sentences. Humans tend to write with greater burstiness, for example, with some longer or complex sentences alongside shorter ones. AI sentences tend to be more uniform. Therefore, when writing the following content I am going to ask you to create, I need it to have a good amount of perplexity and burstiness. Do you understand?
Source: https://youtu.be/Xgc-d7SO4OQ


===== Improve Output =====
It is possible to get chatgpt to improve its own output by having it argue with itself and iterate its replies:
> I want you to act as 2 characters. One is the author who wrote the previous reply. The other is the harsh critic who severely criticizes the weak points of the previous reply. I want you to write the debate of both of those characters as they work together to improve the result. At the end, I want you to give me the final draft.

Another version of this prompt was generated by chatgpt itself:
> Engage in a dialogue as two distinct characters: the first being the author of the previous response, and the second, a stringent critic who meticulously scrutinizes the weaknesses in the author's reply. The critic should provide constructive criticism, highlighting areas for improvement, while the author should respond by refining their original statement based on the feedback. This iterative process should continue until a significantly improved version of the original response is achieved. The dialogue should conclude with the presentation of this final, polished response.