1
Planetdance General / planetdance and AI
« on: September 07, 2024, 12:28:09 PM »
Recently i discovered the use of AI bots for astrological interpretations. I read a text from chatgpt and got enthousiast, so i made the Generate Ai Prompt module (GAP).
My only experience is with chatgpt so far, claude and perplexity by default don't give much text.
On chatgpt i had a really impressive text the first time, in that it combined several factors without me asking for it. I did not give it aspects at that time, only sun, moon, venus and mars in house / sign.
It gave:
Planet in sign
Planet in house
text about that sign / house combo
The next day (model changed?) it gave
Planet in sign / house
So you could argue that the next day it was better because it combined sign/house right away, but it left out the interesting texts about sign / house relations.
Also the second one was much shorter, it omitted several sections that were present in the first attempt: interactions, strengths, challenges and advice. These were really well done if you ask me, it seemed quite intelligent in combining factors. I couldn't get it to use the first approach, asking it why it reported that parameters just get changed, i guess the model was somehow changed?
So i tried adding that as a second question, which did indeed give texts similar to the first attempt, so i made the Header function in GAP.
But i prefer the first approach of giving all that info without asking specifics.
What are your experiences?
My only experience is with chatgpt so far, claude and perplexity by default don't give much text.
On chatgpt i had a really impressive text the first time, in that it combined several factors without me asking for it. I did not give it aspects at that time, only sun, moon, venus and mars in house / sign.
It gave:
Planet in sign
Planet in house
text about that sign / house combo
The next day (model changed?) it gave
Planet in sign / house
So you could argue that the next day it was better because it combined sign/house right away, but it left out the interesting texts about sign / house relations.
Also the second one was much shorter, it omitted several sections that were present in the first attempt: interactions, strengths, challenges and advice. These were really well done if you ask me, it seemed quite intelligent in combining factors. I couldn't get it to use the first approach, asking it why it reported that parameters just get changed, i guess the model was somehow changed?
So i tried adding that as a second question, which did indeed give texts similar to the first attempt, so i made the Header function in GAP.
But i prefer the first approach of giving all that info without asking specifics.
What are your experiences?