Modeling Disclosive Transparency in NLP Application Descriptions

TL;DR

This paper introduces neural language model-based metrics to quantify disclosive transparency in NLP system descriptions, addressing the challenge of its subjective nature and exploring its impact on user understanding.

Contribution

It proposes probabilistic metrics for disclosive transparency that correlate with human judgments and applies them to study transparency's effects in real NLP descriptions.

Findings

Metrics correlate with user and expert opinions.

Transparency influences user perceptions and confusion.

Pilot study demonstrates practical application of metrics.

Abstract

Broader disclosive transparency$-$truth and clarity in communication regarding the function of AI systems$-$is widely considered desirable. Unfortunately, it is a nebulous concept, difficult to both define and quantify. This is problematic, as previous work has demonstrated possible trade-offs and negative consequences to disclosive transparency, such as a confusion effect, where "too much information" clouds a reader's understanding of what a system description means. Disclosive transparency's subjective nature has rendered deep study into these problems and their remedies difficult. To improve this state of affairs, We introduce neural language model-based probabilistic metrics to directly model disclosive transparency, and demonstrate that they correlate with user and expert opinions of system transparency, making them a valid objective proxy. Finally, we demonstrate the use of these metrics in a pilot study quantifying the relationships between transparency, confusion, and user perceptions in a corpus of real NLP system descriptions.