The development of neural machine translation (NMT) has made it possible to realistically test its application to literary texts traditionally reserved for human translation because of their inherent creativity. While recent research (Toral et al. 2020; Oliver González, Toral, and Guerberof Arenas 2019; among others) shows MT’s potential for literary translation, this chapter discusses linguistic issues from the point of view of a literary translator and analyses a section from a contemporary American novel translated into Italian with Google Translate, DeepL, and Microsoft’s Bing Translator. My focus is on how MT deals with those elements that contribute to the creativity and literariness of the text in order to assess the degree of usability of the outputs. The theoretical introduction reviews notions of creativity (Boden 2004; Veale 2015), pragmatics, and cognition (Vega Moreno 2007; Jakobsen 2017) in relation to the characteristics of machine-generated language and MT as highlighted by previous studies (Guerberof Arenas and Toral 2020; Popel 2020; among others). The practice-oriented section analyses the source text—an excerpt from Jackie Polzin’s Brood (2021)—focusing on pragmatic elements and cognitive demands on the reader/translator. Finally, the three machine-translated versions are discussed with particular attention to the latter aspects. To different degrees, the three outputs contain correct and fluent segments, but editing is frequently needed in terms of gender, number, prepositions, tense, lexical choices, information flow, and pragmatic adequacy. Overall, the findings show promising correctness of content and form, but also MT’s current inability to replace situated human experience, which reduces its usefulness for literary translators.