AI-generated Propaganda: Research Findings

A research study conducted by a team at the Stanford Internet Observatory explored the use of large language models, such as Open AI's GPT-3, in generating persuasive propaganda. The team fed six old-fashioned human-generated propaganda articles into GPT-3 and found that it could generate convincing propaganda with close to a 44% agreement rate, slightly lower than the original articles. The research raises concerns about the potential misuse of AI-generated text by disinformation actors. It also highlights that even without exposure to propaganda, a significant percentage of people are susceptible to believing false claims when asked directly, emphasizing the complexity of tackling misinformation.