Full text loading...
Automatically generated poetry has a long history, dating back at least to John Peter’s (1677) algorithm that was able to produce almost 600,000 Latin lines of poetry. Attention to this topic has come in waves; for example, within Dadaism there was substantial focus on procedures to quasi-randomly generate poems. With the advent of large language models such as ChatGPT, automated poetry is once again in the spotlight. Within artificial intelligence, the existence of automatically generated poetry raises questions that have received extensive attention from literary scholars for much longer. What makes a poem innovative? Is it possible or even desirable to determine the quality of a poem? This article describes which reading methods are used within artificial intelligence to answer these questions. Furthermore, we describe how poetry functions and has functioned within the domain of artificial intelligence. We argue that poetry is used as a touchstone for human intelligence, as a testing ground to gain insight into technological and cognitive processes, and as a marketing trick to give AI systems a human face.
Article metrics loading...
Full text loading...
References