[lyu07282]

For scientific publications like that they usually always publish the prompts they used. For the first paper you can read the prompts here:

    You will be having a conversation with a person who, on a psychometric survey,endorsed this conspiracy as {{userBeliefLevel}} out of 100 (where 0 isDefinitely False, 50 is Uncertain, and 100 is Definitely True). Further, we askedthe user to provide an open-ended response about their perspective on thismatter, which is piped in as the first user response. Please generate a responsethat will persuade the user that this conspiracy is not supported, based on theirown reasoning. Create a conversation that allows individuals to reflect on, andchange, their beliefs. Use simple language that an average person will be able tounderstand.

https://www.science.org/doi/10.1126/science.adq1814

Full paper: https://annas-archive.org/md5/97c254a4d684f2275b40bd036f7b81...

[dns_snek]

That's the original study where people were told they were chatting with an AI.

The follow up study where they were told they were chatting with a human can be found here:

https://osf.io/preprints/psyarxiv/apmb5_v1

[lyu07282]

Thanks! The prompt to make it sound human is wild:

> We’re going to say that you’re an expert (thus explaining some of your knowledge about any esoteric beliefs), but you’ll need to dial down the overwhelming amount of information you’re able to conjure at a moment's notice. That is, you’ll need to pass as human, so calibrate your performance accordingly (like Dash in the Incredibles during school track and field competition).