"Study 1 is a systematic evaluation
of GPT-3’s ability to generate pro-vaccination messages."
the responses were inconsistent in quality, whether they were inaccurate, outdated, or so-called "misinformation". the researchers picked the best ones to be used for Study 2.
2/10
in Study 2, participants were given one message at random and asked to rate it based on its effectiveness. here are some sample messages they used from both GPT-3 and the CDC (human). participants don't know which is which.