Yes, exactly I used speech-to-text but actually the chatGPT speech-to-text software on their app because I like the UI better and I think it performs better too. Yeah, the heal heel thing miffed me slightly but I think it is a fun artifact since it doesn’t actually change the meaning.
Well for one I didn’t prompt for a whole essay. In one chat I lightly edited the snippets from my walk, then I took the final essay generated from another chat about the Black Chess Box to synthesise into the Sidebar and similarly for a different conversation again for part 2 and then finally, which is where Claude has the advantage—because at this point the context would be too large for ChatGPT 4o for instance—you just ask for either a brief or extended conclusion to all discussed in the chat. In summary, having separate conversations to develop sections and bring them all together in one final chat. This worked well for this essay because the progression from section to section didn’t need to be that strong but idk what one would do if that were the case.
I have tried other methods in the past and in general, there’s no one size fits all (for instance sometimes the project function can allow you to tackle reports over 10 pages long, then sometimes it just gets stuck in loops.) The best thing to do is try to leverage the advantages you have and experiment.
Hi, nim!
Thanks for commenting : )
Yes, exactly I used speech-to-text but actually the chatGPT speech-to-text software on their app because I like the UI better and I think it performs better too. Yeah, the heal heel thing miffed me slightly but I think it is a fun artifact since it doesn’t actually change the meaning.
Well for one I didn’t prompt for a whole essay. In one chat I lightly edited the snippets from my walk, then I took the final essay generated from another chat about the Black Chess Box to synthesise into the Sidebar and similarly for a different conversation again for part 2 and then finally, which is where Claude has the advantage—because at this point the context would be too large for ChatGPT 4o for instance—you just ask for either a brief or extended conclusion to all discussed in the chat. In summary, having separate conversations to develop sections and bring them all together in one final chat. This worked well for this essay because the progression from section to section didn’t need to be that strong but idk what one would do if that were the case.
I have tried other methods in the past and in general, there’s no one size fits all (for instance sometimes the project function can allow you to tackle reports over 10 pages long, then sometimes it just gets stuck in loops.) The best thing to do is try to leverage the advantages you have and experiment.
Anyway I hope that answers your question
Matthew