New study out today! We had a bunch of Democrats and Republicans do an Ideological Turing Test. People either portrayed themselves or pretended to be from the opposite party, and then another group tried to pick out the real partisans from the fake ones:
Conversation
We randomly assigned 902 Republicans and Democrats to write at least 100 words based on one of two prompts: either "I'm a DEMOCRAT because..." or "I'm a REPUBLICAN because..." So half were telling the truth, and half were faking. (This was pre-ChatGPT)
Then we got another sample of 746 partisans and had them try to tell the difference between the real statements and the fake ones. Both sides got bonuses for good performance.
Before you see the results, you can try this study for yourself! See if you can tell the difference between a real partisan and an enemy in disguise: ituringtest.com
Results: Democrats could not tell the difference between real Democrats and undercover Republicans
And Republicans could not tell the difference between real Republicans and undercover Democrats:
People did no better than chance, and they were equally bad for statements from their own party and the opposite party:
So both Democrats and Republicans can pass an Ideological Turing Test! Full paper and tons more results here:
![David Watson 🥑](data:image/svg+xml,<svg xmlns="http://www.w3.org/2000/svg" width="96" height="96"><rect fill-opacity="0"/></svg>)
Post your reply
And here's the PsyArxiv version for all you PDF-heads out there: osf.io/preprints/psya
This is cool, my next question would be whether NIMBYs, sorry, "neighborhood character protectors" can pass a Turing test vs YIMBYs
America's biggest DEI hire's dark past is coming back to haunt her. For being one of the government’s most prolific & boisterous chucklers, Kamala Harris has quietly proven herself to be another unwelcome Face of Tyranny.
Rate proposed Community Notes
I think your limitations section is quite limited. This test measures whether people can write a statement that people think someone from the other party would write - it doesn't measure what people think the other party actually believes.
Apologies for responding to the thread and not the article, but: does this conclusion of “passing” depend at all on how confident people were in their guesses? Like, does it matter whether they believed the fake Democrat was real or just had no idea but guessed?
What about independents? I identified 9 out of 10 and the one I missed I was pretty unsure of. It was a real Republican that thought was fake. Very stereo typical answer,?I thought maybe too much so.
This ITT is too easy because it’s just describing labeled beliefs
A real ITT is understanding justifications for those beliefs
Most people's writing ability is quite poor. Those writing about their real beliefs probably didn't write all that well to start with, so even if the fakers don't really understand the other side, they'll still blend in fine, since the other side wasn't communicating well anyway.
Wait wouldn’t passing the test mean recognizing the impostors?
