AI humanizers and the crisis of information integrity: implications for scientific writing

Abstract

AI humanizers are tools designed to disguise AI-generated text as human-written. These tools have triggered a crisis in information integrity. This paper examines the reasons behind the rise of AI humanizers and their damaging effects on scientific writing. The primary reasons for adopting AI humanizers stem from institutional and psychological pressures. Students and researchers face a publish-or-perish culture and rigid academic policies that impose arbitrary AI-percentage thresholds. These tools are increasingly used as compliance mechanisms to avoid being flagged by AI detection. Furthermore, a lack of confidence in information literacy and the pressure of time poverty drive users to seek an information bypass, prioritizing efficient yet shallow output over deep cognitive engagement. The effects of this reliance are ethically and educationally corrosive. By decoupling writing from thinking, AI humanizers produce pseudo-information that mimics scholarly fluency but lacks epistemic grounding. This leads to a decline in critical information processing. Long-term consequences include the stunting of information literacy, the rise of learned helplessness, and the systemic pollution of institutional repositories with content lacking authentic inquiry. The paper argues that information integrity cannot be recovered through technical policing alone; it requires a return to process-oriented assessments that prioritize information provenance and the authentic knowledge engagement.