•  
  •  

Home > CIIMA > Vol. 24 (2026) > Iss. 1

 

Communications of the IIMA

Abstract

The rapid proliferation of generative AI tools has fundamentally altered the volume, velocity, and perceived quality of information produced across academic and professional settings (Bommasani et al., 2021; Stanford Institute for Human-Centered Artificial Intelligence [HAI], 2025). Yet this expansion has introduced a critical and underexamined threat to sustainable information ecosystems: what this paper formally terms AI slop ; defined as AI-generated content that achieves surface-level plausibility while offering limited substantive value, accuracy, or contextual depth. At scale, AI slop degrades information quality and erodes institutional trust (Bender et al., 2021; Weidinger et al., 2021). More critically, it generates what this paper identifies as invisible verification labor , the unacknowledged cognitive work of reviewing, filtering, correcting, and validating AI output that is silently redistributed onto human recipients. A recent study from Stanford University and BetterUp Labs underscores the urgency of this problem: approximately 40% of knowledge workers reported encountering low-effort AI-generated output within a single month, an experience that produced measurable rework, decision fatigue, and diminished trust in AI-assisted workflows (Niederhoffer et al., 2025). To address this gap, this paper employs a systematic literature review of emerging research on AI output quality, cognitive load, information trust, and human-centered system design across academic and professional organizational contexts. We develop a three-dimensional taxonomy of AI slop — spanning intent (accidental, productivity-driven, or deceptive), detectability (obvious, semi-hidden, or indistinguishable), and organizational impact (low-risk annoyance to high-risk institutional harm) ; positioning this taxonomy as a conceptual anchor for future empirical research and governance design. Our analysis argues that AI slop is not a content quality edge case but a structural byproduct of deploying generative AI without quality accountability frameworks (Bommasani et al., 2021). Left unaddressed, it progressively undermines the cognitive sustainability, informational sustainability, and organizational sustainability of the systems it inhabits. This paper contributes a human-centered mitigation framework that treats information quality and the equitable distribution of verification labor as shared institutional responsibilities — advancing sustainable, ethical, and future-oriented information management in the generative AI era.

Download

Share

COinS