[论文解读] AI Transparency in the Age of LLMs: A Human-Centered Research Roadmap

Q. Vera Liao, Jennifer Wortman Vaughan|arXiv (Cornell University)|Jun 2, 2023

Topic Modeling被引用 25

一句话总结

本文主张以人为本的透明度路线图用于大语言模型（LLMs），概述挑战、利益相关者需求，以及四种常见的透明度方法（模型报告、评估结果、解释和不确定性沟通），并提供对将它们应用于LLMs的指导。

ABSTRACT

The rise of powerful large language models (LLMs) brings about tremendous opportunities for innovation but also looming risks for individuals and society at large. We have reached a pivotal moment for ensuring that LLMs and LLM-infused applications are developed and deployed responsibly. However, a central pillar of responsible AI -- transparency -- is largely missing from the current discourse around LLMs. It is paramount to pursue new approaches to provide transparency for LLMs, and years of research at the intersection of AI and human-computer interaction (HCI) highlight that we must do so with a human-centered perspective: Transparency is fundamentally about supporting appropriate human understanding, and this understanding is sought by different stakeholders with different goals in different contexts. In this new era of LLMs, we must develop and design approaches to transparency by considering the needs of stakeholders in the emerging LLM ecosystem, the novel types of LLM-infused applications being built, and the new usage patterns and challenges around LLMs, all while building on lessons learned about how people process, interact with, and make use of information. We reflect on the unique challenges that arise in providing transparency for LLMs, along with lessons learned from HCI and responsible AI research that has taken a human-centered perspective on AI transparency. We then lay out four common approaches that the community has taken to achieve transparency -- model reporting, publishing evaluation results, providing explanations, and communicating uncertainty -- and call out open questions around how these approaches may or may not be applied to LLMs. We hope this provides a starting point for discussion and a useful roadmap for future research.

研究动机与目标

• 识别由大型语言模型（LLMs）及其生态系统带来的独特透明度挑战。
• 倡导以人为本的视角，考虑多样化的利益相关者与情境。
• 回顾四种常见的透明度方法，并讨论它们在LLMs中的适用性。
• 提出方向和开放性问题，以指导未来的LLM透明度研究。

提出的方法

• 综合来自人机交互（HCI）与负责任AI的现有AI透明度研究，以绘制利益相关者的目标与需求。
• 描述LLM特定的透明度挑战，如能力不可预测性和不透明架构。
• 概述四种常见的透明度方法，并分析它们在LLMs中的潜在应用（模型报告、发布评估结果、解释、不确定性沟通）。
• 讨论对界面设计、心智模型、信任和用户可操作性的影响。

实验结果

研究问题

• RQ1 是什么使得LLMs的透明度与较小模型的透明度不同？
• RQ2 如何将模型报告、评估、解释和不确定性沟通改编以支持LLMs生态系统中的利益相关者？
• RQ3 需要哪些设计与评估方面的考量，以使透明度与用户的目标和心智模型对齐？
• RQ4 在与LLM透明度相关的负责任AI治理与政策方面，尚存哪些未解的开放性问题？

主要发现

• 透明度应为不同利益相关者和情境提供适当的理解，而不仅仅是提供信息。
• LLMs带来挑战，如能力不可预测性、庞大且不透明的架构，以及广泛且在发展中的利益相关者群体。
• 以人为本的视角强调目标与任务，并将透明度与实现下游活动（如调试、决策和治理）联系起来。
• 沟通形式与设计（可视化、自然语言、框架）在很大程度上影响透明信息的处理与使用。