ChatGPT is probably a future Nobel Prize winner in literature.
New research, published in the journal Scientific Reports, demonstrates the inability of readers to distinguish between poems created by artificial intelligence and humans. The majority of participants in the experiment preferred the AI poems.
Researchers Brian Porter and Edward Maceri from the Department of History and Philosophy of Science at the University of Pittsburgh tested the ability of 1634 participants to distinguish between poetry created by AI and that written by a human poet.
Participants were randomly presented with ten poems: five written by famous poets, including William Shakespeare, Lord Byron, Emily Dickinson, and Thomas Eliot, and five poems generated by ChatGPT3.5 in the style of these poets. Participants were more likely to believe that the AI poems were written by humans, and the five poems they thought were least likely to be human were by real poets.
In the second experiment, another group of 696 participants rated poems on 14 characteristics, such as quality, beauty, emotion, rhythm, and originality. The participants were randomly divided into three groups, who were told that the poems were written by humans, created by AI, or were not given any information about the authorship.
Participants who knew about the authorship of the poems rated them lower on all criteria. Instead, the group without information about the origin of the works preferred the generated poems.
The researchers explain the results by the simplicity and accessibility of AI poems compared to the works of classics. Readers mistook this ease of understanding as a sign of human authorship.
Source: phys