Understanding and Perception of Automated Text Generation among the Public: Two Surveys with Representative Samples in Germany

Abstract

The application of automated text generation (ATG) technology has evolved rapidly in the last several years, leading to more available content being produced by artificial intelligence (AI). With the release of ChatGPT in November 2022, a tool is now at hand that allows virtually everyone to create naturally sounding text on any topic. However, research on public beliefs about and attitudes toward ATG is lacking. Two surveys with representative samples (n1 = 1,028, n2 = 1,013) examined the German population’s concepts of and attitudes toward AI authorship in 2022 and 2023. The results revealed a preference for human authorship across a wide range of topics and a lack of knowledge concerning the function, data sources, and responsibilities of ATG. Using multiple regression analysis with k-fold cross-validation, we identified several significant predictors of the intention to read AI-written texts in the future. Despite the release of ChatGPT, we observed stability across most variables and minor differences between the two survey points regarding concepts about ATG. We discuss the findings against the backdrop of the ever-increasing availability of automated content and the need for an intensive societal debate about the chances and limitations of this technology.

The Effects of Assumed AI vs. Human Authorship on the Perception of a GPT-generated Text

“…The study was conducted in January 2022, when peoples' experience with and attitude toward ATG and NLG might not have been highly developed due to missing labeling requirements. In addition, there is evidence that people had no clear concepts regarding ATG and that this has not changed considerably since the release of ChatGPT 17,18 . Against the backdrop that it was a new experience for participants that an AI was able to write about a scientific topic to this extent, the findings are highly interesting.…”