According to a new report from a team at Stanford, GPT-3 can influence humans’ opinions on policy issues. AI-generated messages were found to be as persuasive as arguments from regular people. They were also perceived as more logical and factual than human messages.
While this raises questions about regulating AI applications in political contexts to prevent misinformation, it also underlines the power of AI. If GPT-3 is more persuasive than the everyday person, we can perhaps learn even more from it than expected.
Read this study at OSF.
コメント