Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Understanding Stakeholders' Perceptions and Needs Across the LLM Supply Chain

Agathe Balayn, Lorenzo Corti, Fanny Rancourt, Fabio Casati, Ujwal Gadiraju

TL;DR

The paper addresses the problem that explainability and transparency in LLMs are often studied from narrow stakeholder perspectives. It uses a qualitative study with 71 stakeholders across organizations to characterize who needs XAI/TAI information, what information is needed, and why. The findings reveal diverse information needs, cross-functional purposes, and challenges such as knowledge gaps, confidentiality concerns, and rapid model updates, highlighting alignment and expansion of prior frameworks like the foundation model transparency index. The authors propose a supply-chain grounded conceptual framework and methodological guidance to better align XAI/TAI with real-world roles and governance, with implications for policy, accountability, and responsible design across the LLM supply chain.

Abstract

Explainability and transparency of AI systems are undeniably important, leading to several research studies and tools addressing them. Existing works fall short of accounting for the diverse stakeholders of the AI supply chain who may differ in their needs and consideration of the facets of explainability and transparency. In this paper, we argue for the need to revisit the inquiries of these vital constructs in the context of LLMs. To this end, we report on a qualitative study with 71 different stakeholders, where we explore the prevalent perceptions and needs around these concepts. This study not only confirms the importance of exploring the ``who'' in XAI and transparency for LLMs, but also reflects on best practices to do so while surfacing the often forgotten stakeholders and their information needs. Our insights suggest that researchers and practitioners should simultaneously clarify the ``who'' in considerations of explainability and transparency, the ``what'' in the information needs, and ``why'' they are needed to ensure responsible design and development across the LLM supply chain.

Understanding Stakeholders' Perceptions and Needs Across the LLM Supply Chain

TL;DR

The paper addresses the problem that explainability and transparency in LLMs are often studied from narrow stakeholder perspectives. It uses a qualitative study with 71 stakeholders across organizations to characterize who needs XAI/TAI information, what information is needed, and why. The findings reveal diverse information needs, cross-functional purposes, and challenges such as knowledge gaps, confidentiality concerns, and rapid model updates, highlighting alignment and expansion of prior frameworks like the foundation model transparency index. The authors propose a supply-chain grounded conceptual framework and methodological guidance to better align XAI/TAI with real-world roles and governance, with implications for policy, accountability, and responsible design across the LLM supply chain.

Abstract

Explainability and transparency of AI systems are undeniably important, leading to several research studies and tools addressing them. Existing works fall short of accounting for the diverse stakeholders of the AI supply chain who may differ in their needs and consideration of the facets of explainability and transparency. In this paper, we argue for the need to revisit the inquiries of these vital constructs in the context of LLMs. To this end, we report on a qualitative study with 71 different stakeholders, where we explore the prevalent perceptions and needs around these concepts. This study not only confirms the importance of exploring the ``who'' in XAI and transparency for LLMs, but also reflects on best practices to do so while surfacing the often forgotten stakeholders and their information needs. Our insights suggest that researchers and practitioners should simultaneously clarify the ``who'' in considerations of explainability and transparency, the ``what'' in the information needs, and ``why'' they are needed to ensure responsible design and development across the LLM supply chain.
Paper Structure (4 sections, 1 figure)

This paper contains 4 sections, 1 figure.

Table of Contents

  1. A Case for XAI Across the Supply Chain
  2. Study Methodology
  3. Results
  4. Discussion & Implications

Figures (1)

  • Figure 1: Information needs per category of stakeholders (fraction of the participants who mentioned the need). We excluded information needs mentioned by a single category of stakeholders: developers - LLM confidence (14%); customer-centered roles - log of LLM inputs/outputs (18%), fairness performance (9%); UX researcher - open-source (25%), work of other teams (50%); end-user - LLM creators (100%).