Yesterday I asked GPT to generate a complex prompt for me. I noticed the prompt did not have any special brackets [ ] or pound # symbols nothing but clear plain English. At that point I noticed something changed about prompts.
So, maybe the prompt engineering might become more complex or effective. Because you wrote in text that approaches are from papers, and new papers are published all the time. I think prompt engineering will be more specific until we have more nothing to discover about that.
Very interesting and very clear ! Just a thought: the advantage of the prompt is that it asks the user to clarify his thoughts, to try to structure his request and his thoughts so that the answer is as relevant as possible. With the arrival of models that ‘reason’, there's a significant risk that users will become increasingly lazy to think... Not a very exciting idea, is it?
Yesterday I asked GPT to generate a complex prompt for me. I noticed the prompt did not have any special brackets [ ] or pound # symbols nothing but clear plain English. At that point I noticed something changed about prompts.
So, maybe the prompt engineering might become more complex or effective. Because you wrote in text that approaches are from papers, and new papers are published all the time. I think prompt engineering will be more specific until we have more nothing to discover about that.
Excellent article and I think you’re absolutely right 🙂👍
Very interesting and very clear ! Just a thought: the advantage of the prompt is that it asks the user to clarify his thoughts, to try to structure his request and his thoughts so that the answer is as relevant as possible. With the arrival of models that ‘reason’, there's a significant risk that users will become increasingly lazy to think... Not a very exciting idea, is it?
Thank you, it’s always interesting and informative to read you. I started your course "
ChatGPT Unleashed: The Complete ChatGPT Course 2024
" on Udemy