Hello everyone, trying to train ChatGPT to be silent and it seems to work, despite some inaccuracies on my side (I’m not a native English speaker).
Is it likely that if this training was run by thousands of users, ChatGPT would start to adopt this behavior or it’s just science-fiction? 
Just out of curiosity, but if it went viral it would be a nice experiment.
1 Like
ChatGPT does not save your data. You are not training it. You use it.
Everything you want to achieve can be done with one prompt.
When you have a chat with multiple prompts. That is called learning.
Welcome to the developer community.
1 Like
Hi, thank you for your kind response. Based on this, it seems that the only possible approach would be to create a combination of prompts that generates this behavior (silence), but, if I understood correctly, this process would need to be repeated each time.
An AI Model has no brain. It may appear intelligent but it is not.
Let me try to give you an analogy.
Imagine a machine that has 5 buttons and each plays a random melody.
You pressed one button and you like the first part of the melody. You tell it to the owner of the machine who then changes the machine to produce more similar music.
This does not make the machine a composer and you can’t change the machine either.
The owners of the machine only change it when a lot of button clickers rate some type of music good.
“Prompt injection” is the word to beginn your learning path.
Read about it. Watch ten hours of videos about it.
You are welcome.
Hi, thanks for the explanations and suggestions. I’ll definitely look into it more 
PS: a funny coincidence, I came across this message this morning. Deepseek side effects? 
