The Transparency Laboratory: A Comparative Framework for Mediated Authenticity in Chemical and Algorithmic Disclosure

Abstract

The proliferation of generative AI and algorithmic curation has created an authenticity crisis, as audiences struggle to distinguish genuine human expression from computationally optimized performance. This study challenges the assumption that algorithmic mediation is categorically unlike previous mediated communication by comparing it with Korean celebrity “drunk talk shows” on YouTube, where alcohol mediates on-camera disclosure. Integrating Goffman’s performance theory, Spence’s signaling theory, and Bishop’s algorithmic folk theories, the study introduces the transparency laboratory: a methodological logic using overt, consented mediation to reveal strategic dynamics concealed by opaque, non-consented algorithmic mediation. A qualitative, illustrative analysis of five purposively selected episodes grounds the framework without claiming representativeness or statistical evidence. The study also proposes a five-stage model of algorithmic genre normalization as a testable theoretical framework, rather than an established empirical finding. Its contributions are threefold: theoretically, the transparency laboratory offers a transferable methodological logic; conceptually, authenticity is reframed as shaped by optimization logics rather than a binary property of expression; practically, governance should target mediating-agent incentives rather than individual authenticity