تقرير
Supporting Human Raters with the Detection of Harmful Content using Large Language Models
| العنوان: | Supporting Human Raters with the Detection of Harmful Content using Large Language Models |
|---|---|
| المؤلفون: | Thomas, Kurt, Kelley, Patrick Gage, Tao, David, Meiklejohn, Sarah, Vallis, Owen, Tan, Shunwen, Bratanič, Blaž, Ferreira, Felipe Tiengo, Eranti, Vijay Kumar, Bursztein, Elie |
| سنة النشر: | 2024 |
| المجموعة: | Computer Science |
| مصطلحات موضوعية: | Computer Science - Cryptography and Security |
| الوصف: | In this paper, we explore the feasibility of leveraging large language models (LLMs) to automate or otherwise assist human raters with identifying harmful content including hate speech, harassment, violent extremism, and election misinformation. Using a dataset of 50,000 comments, we demonstrate that LLMs can achieve 90% accuracy when compared to human verdicts. We explore how to best leverage these capabilities, proposing five design patterns that integrate LLMs with human rating, such as pre-filtering non-violative content, detecting potential errors in human rating, or surfacing critical context to support human rating. We outline how to support all of these design patterns using a single, optimized prompt. Beyond these synthetic experiments, we share how piloting our proposed techniques in a real-world review queue yielded a 41.5% improvement in optimizing available human rater capacity, and a 9--11% increase (absolute) in precision and recall for detecting violative content. |
| نوع الوثيقة: | Working Paper |
| URL الوصول: | http://arxiv.org/abs/2406.12800 |
| رقم الأكسشن: | edsarx.2406.12800 |
| قاعدة البيانات: | arXiv |
كن أول من يترك تعليقا!