Thanks for giving it a shot. Unfortunately I think there may be some subtle things in the modifications that could influence the rest of the output: for instance, there's a grammatical error in "The presentation's clever title and subtitle was"
And to reiterate "charming and witty" so soon after the initial "delightful/sophisticated/clever" gives the message a different tenor (I may have already overdone it with the qualifiers myself).
I think this is a subtle, important aspect of the problem: it's not quite just attaching positive qualifiers—they also have to be strung together in a style reminiscent of the contexts in which those qualifiers would actually be used as legitimate descriptions of quality material.
In any case, still some pretty coherent interesting output. I need to find a way to play with this on my own though (I jumped into aidungeon.io for a minute, but didn't see a good way of setting up the prompt. It is using GPT-3?)
And to reiterate "charming and witty" so soon after the initial "delightful/sophisticated/clever" gives the message a different tenor (I may have already overdone it with the qualifiers myself).
I think this is a subtle, important aspect of the problem: it's not quite just attaching positive qualifiers—they also have to be strung together in a style reminiscent of the contexts in which those qualifiers would actually be used as legitimate descriptions of quality material.
In any case, still some pretty coherent interesting output. I need to find a way to play with this on my own though (I jumped into aidungeon.io for a minute, but didn't see a good way of setting up the prompt. It is using GPT-3?)