“It’s giving AI”: Reading ambiguously-authored texts and the role of felt sense

To understand how human readers navigate a literate landscape that newly includes AI-generated prose, we asked participants (n=76) to read and make decisions about who and/or what is responsible for writing anonymized, “ambiguously-authored” texts. Findings suggest that readers’ assumptions about who and/or what wrote a text are rooted in “felt sense.” Prompting participants to make their “felt sense” explicit allowed us to catalog the evidential warrants participants relied on when making authorship decisions. Enabled by a modified grounded theory approach to analysis, we constructed two main themes. First, readers are “triggered” by certain textual cues that, when combined with prior experiences and knowledge, evidentially warrant assumptions about who and/or what wrote a text. Second, after recognizing the consequences of making one’s felt sense explicit, some readers experience what we call an “axiological crisis.” Axiological crises emerge when participants meta-cognitively hear or see themselves attributing certain characteristics and values to an AI text-generator or human author. We conclude by reimagining the axiological crisis as an opportunity for improving metacognitive awareness about how felt sense affects our reading practices.