Thanks to ChatGPT and other AI tools, we, the readers, sometimes don’t even know if the texts were written by an AI or if an author, in this case me (Stefan Kny), really wrote it. Can the AI reveal itself?
To answer this question, we ask Mindverse AI. I insert the answers 1:1 via copy and paste.
Questions: How do you recognize texts written by an AI?
It is sometimes difficult to recognize texts written by an AI. However, there are some indications that a text is created by an AI. Some of the most common features are: 1) The text usually contains little or no emotion or feeling. 2) The grammar of the text is sometimes not quite correct. 3) The text can be written in a very general style and often lacks specific details. 4) The text can be very systematic and often written in a predictable pattern. 5) The text can be very repetitive and repeat the same information several times in different ways. If you notice several of these characteristics when reading a text, it is very likely that the text was created by an AI.
Now we know that the reply text is from an AI, without any interference. If we now assume simple copy and paste, we will notice that the response is not nicely structured.
So I ask Mindverse AI: please structure the answers for me.
1) The text usually contains little or no emotion or feeling: This means that the text does not contain many descriptions or comparisons designed to evoke an emotional or sentimental response in the reader.
2) The grammar of the text is sometimes not quite correct: the text may contain ungrammatical sentences, bad sentence order or wrong choice of words.
3) The text can be written in a very general style and often lacks specific details: this means that the text provides little information that belongs to a specific context. He may also have a very general perspective that is not tailored to a specific subject.
4) The text can be very systematic and often written in a predictable pattern: the text usually follows a certain structure and proceeds in a predictable order.
5) The text can be very repetitive and repeat the same information several times in different ways: the author often repeats single words or phrases in the text to ensure that the message is clearly understood.
For the past few months, I have been working intensively with AI tools. I will keep you updated on this in the near future. I will set AI different tasks. Where are their strengths, where are their weaknesses, and how do you best use AI for yourself.
My personal answer to the question:
In its response, the AI often repeats the question or paraphrases it (point 5). It reminds me a bit of my school days. Here, too, the students asked by the teacher repeated the question completely in an answer, mostly to get some time to think about the answer or to get the neighbor to say it to them. In other words, almost human behavior. Of course, the point of human error does not appear in an AI, ie. there are no gross spelling errors with an AI, but if a text goes through a proofreading department several times, it must also be error-free. As you can see, the topic is both complex and exciting and will be addressed in more detail in the future.
More on the subject of AI soon.
Photo: AI-generated photo by Mindverse
How do you recognize texts written by an AI?
Here you can find the current live stream episodes. More from Web3, NFT and Metaverse.