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

Area of Science:

• Human-Computer Interaction
• Artificial Intelligence Ethics
• Societal Impact of AI

Background:

• Automated text generation (ATG) and AI authorship are increasingly prevalent, with tools like ChatGPT enabling widespread content creation.
• Understanding public perception of AI-generated content is crucial for optimizing its use and managing societal integration.
• Limited research exists on public attitudes and beliefs regarding ATG and AI authorship.

Purpose of the Study:

• To investigate the German population's concepts and attitudes toward AI authorship.
• To identify predictors of the intention to read AI-written texts.
• To assess changes in perceptions of ATG between March 2022 and July 2023.

Main Methods:

• Two representative surveys conducted in Germany (March 2022, n=1028; July 2023, n=1013).
• Analysis included multiple regression with k-fold cross-validation to identify significant predictors.
• Data collection focused on concepts, attitudes, and intentions related to AI authorship and ATG.

Main Results:

• A strong preference for human authorship was observed across various topics.
• Public knowledge regarding ATG's function, data sources, and responsibilities was found to be lacking.
• Attitudes toward using ATG, performance expectancy, general AI attitudes, and specific attitudes toward ChatGPT/ATG predicted the intention to read AI-generated texts.
• Perceptions of ATG remained largely stable between the two survey periods, despite the release of ChatGPT.

Conclusions:

• The study highlights a societal preference for human-authored content and a gap in public understanding of ATG.
• Findings underscore the need for a broader societal discussion on the implications of AI-generated content.
• Continued research is essential to navigate the increasing presence of automated content and its societal impact.