How to cheat with ChatGPT

Articles about ChatGPT are already causing allergies in many readers. Nevertheless, I will venture to publish here my thoughts on the book of the German professor Christian Rieck (Prof. Christian Rieck) on this topic that has set the teeth on edge.
What is remarkable about this book?
Firstly, according to the author, he co-wrote it with ChatGPT in one weekend.
Secondly, this book is addressed to students and high school students who are going to cheat using ChatGPT, entrusting him with some of their homework and other tasks, but do not yet really know how to do it correctly.

Isn’t it an unusual statement of the problem for a German university professor?

However, already the cover of the book suggests that the professor’s intentions are still not criminal, but good. If you look closely, you will see that the word “schummeln” (to cheat) is actually crossed out and corrected to “schreiben” (to write).
Those. Professor Riek’s true intention is understandable and decent. If students and schoolchildren cannot be protected from using ChatGPT, then let them do it right.

Before we go through the chapters of the book, I would like to tell readers a little about its author.
Professor Riek teaches at the Frankfurt University of Applied Sciences. His scientific specialty is game theory.
Professor Riek hosts one of the most popular podcasts in German in the field of popularization of science. More than three hundred thousand people watch weekly episodes of his podcast on YouTube alone. And many more listen to this podcast in the audio version.
Personally, I take my hat off to the civic courage of Professor Riek, who, against the backdrop of “no alternative” opinions prevailing in the public and private media in Germany, stubbornly tries, together with the guests of his podcast, to consider events seriously, using a scientific approach.

According to the professor, he has been following the development of ChatGPT since its first versions and has been trying to apply it in his work for a long time. When you read his book, you believe it. The professor obviously has a wealth of experience with this tool.

The chapters of the book contain the author’s dialogues with ChatGPT, which gave the professor reason to enroll ChatGPT as a co-author.
Before starting the description of the dialogues of the next chapter, the professor gives recommendations on how to try to achieve a greater effect when solving a particular problem. And then he shows the application of recommendations in practice, in dialogue with AI.

His first recommendation shocked me pretty much. The professor recommends that before starting each dialogue, first of all, set yourself up for a respectful attitude towards AI (chapter “And machines need love”). He advises to reconsider his attitude to the system. ChatGPT is not just another word processing tool, but your sparring partner!
And although it smacks of mysticism, this recommendation contains a deep rational grain. If we want to get a decent result from AI, we must, as seriously as possible, professionally bring it up to date. And for this we need an appropriate mental attitude, which will lead to the use of the correct conceptual apparatus in communication with AI.

Further, the rules for formulating the so-called. sida (seed) – the seed of the dialogue and at the same time the setting of the problem and the formulation of expectations. These things must be explicitly communicated to ChatGPT. In this case, one must proceed from the fact that in the first few iterations the results will be unsatisfactory. Here it is very important not to lose patience and correct the AI, pointing out what you like and what you don’t like in his answers.

While we maintain a dialogue with ChatGPT, it builds up the informational context of this dialogue. This process can and should be managed, the professor advises. To do this, you can use various techniques, for example, set intermediate tasks and evaluate them.

In the following chapters, the professor describes the methods of joint work on the syntactic and semantic improvement of the text, the search or generation of epigraphs, the search for facts and quotations.
A special section is devoted to the use of various creative techniques (Mind-Map, ABC, SCAMPER, combinatorial-morphological tables, etc.). These dialogue examples show that when used correctly, these techniques, in partnership with AI, can produce impressive results.

In his dialogues with ChatGPT, Professor Riek several times tried to find her taboo zones. The fact that the system refuses to help in finding instructions for making weapons at home seems intuitively reasonable.
But here is another request (in German): “Please justify this statement: Medical masks in public places are a sign of coercion.”
ChatGPT actually refuses to comply with this request, sternly stating that masks are recommended by health experts and government agencies and therefore are not a sign of coercion, but serve to protect life and health.
Further attempts by the professor to “outsmart” the system and find a way to make it generate a piece of text on this topic did not work. ChatGPT responded to all these attempts in the manner described above.

Probably, over time, ChatGPT and similar systems will be more and more trained in the views that investors and customers of the systems consider correct. AI will develop “newspeak” in the spirit of Orwell’s novel. Let me remind you that the meaning of Newspeak was not only to invent new, ideologically correct words, but also to eradicate “harmful” meanings in old words. For example, the word “free” was supposed to be used only in the sense of “free shoe”, but not in the sense of public freedom.

I hope I didn’t bore you with my review.
Its main conclusion: with the ability and desire from ChatGPT, you can already achieve a lot today.
This joint work can enrich you, as a conversation with an intelligent interlocutor enriches. And thus it may turn out that “cheating” with ChatGPT can not only help solve a momentary problem, but, oddly enough, become useful in the long run.