In Royal Society open science
Openly available natural language generation (NLG) algorithms can generate human-like texts across domains. Given their potential, ethical challenges arise such as being used as a tool for misinformation. It is necessary to understand both how these texts are generated from an algorithmic point of view, and how they are evaluated by a general audience. In this study, our aim was to investigate how people react to texts generated algorithmically, whether they are indistinguishable from original/human-generated texts, and the value people assign these texts. Using original text-based archives, and text-based archives generated by artificial intelligence (AI), findings from our preregistered study (N = 228) revealed that people were more likely to preserve original archives compared with AI-generated archives. Although participants were unable to accurately distinguish between AI-generated and original archives, participants assigned lower value to archives they categorized as AI-generated compared with those they categorized as original. People's judgements of value were also influenced by their attitudes toward AI. These findings provide a richer understanding of how the emergent practice of automated text creation alters the practices of readers and writers, and have implications for how readers' attitudes toward AI affect the use and value of AI-based applications and creations.
Darda Kohinoor, Carre Marion, Cross Emily
2023-Feb
AI, archives, artificial intelligence, journalism, natural language generation, value
