Built independently by an author, for readers. Read the story and support ChapterPal
ChapterPal Logo
Log inSign up

keyword

research reproducibility

Research reproducibility is the extent to which independent investigators can achieve consistent findings and conclusions when applying the same scientific methodologies, experimental designs, or computational workflows. In computational and empirical sciences, it fundamentally depends on the transparency, accessibility, and stability of data, code, software environments, and measurement tools. When research relies on evolving software dependencies, dynamic external tools, or undocumented analytical steps, subsequent attempts to evaluate or compare outcomes can yield divergent results. Consequently, research reproducibility serves as a core standard of scientific rigor by ensuring that published findings are auditable, verifiable, and resilient over time across different research settings.

1 item

On the Challenges of Using Black-Box APIs for Toxicity Evaluation in Research

On the Challenges of Using Black-Box APIs for Toxicity Evaluation in Research

Luiza Pozzobon, Beyza Ermis, Patrick Lewis, Sara Hooker

OrganizationsCohereRecod.aiSchool of Electrical and Computer EngineeringUniversity of Campinas

Why you should read this

Reveals how unannounced updates to commercial toxicity evaluation APIs like Perspective alter benchmark leaderboards and invalidate prior research conclusions, while establishing concrete guidelines to ensure reproducible evaluations over time.

Perception of toxicity evolves over time and often differs between geographies and cultural backgrounds. Similarly, black-box commercially available APIs for detecting toxicity, such as the Perspective API, are not static, but frequently retrained to address any unattended weaknesses and biases. We evaluate the implications of these changes on the reproducibility of findings that compare the relative merits of models and methods that aim to curb toxicity. Our findings suggest that research that relied on inherited automatic toxicity scores to compare models and techniques may have resulted in inaccurate findings. Rescoring all models from HELM, a widely respected living benchmark, for toxicity with the recent version of the API led to a different ranking of widely used foundation models. We suggest caution in applying apples-to-apples comparisons between studies and lay recommendations for a more structured approach to evaluating toxicity over time. 1

Added

2026-09-26