Scholarlink

<p>Responsible Artificial Intelligence (RAI) has emerged as a critical concern as organizations increasingly adopt powerful AI technologies without ensuring the development of organizational capabilities required to embed strong ethical practices into AI workflows at scale. In the absence of such capabilities, RAI may become symbolic rather than structural, with firms falling into ethics washing or governance gaps despite good intentions. This study investigates what mix AI assets and capabilities support strong RAI practices. Using longitudinal survey data from 193 AI-active large multinational firms between 2018 and 2024, we identify three distinct RAI adoption trajectories via Latent Class Analysis. We then use Random Forest machine learning to evaluate the best bundle of AI-related assets and capabilities in shaping these trajectories. If RAI practices adoption can be driven by a desire for regulatory compliance, our results clearly show that this is the orchestration of three capabilities, -AI talent managment, ecosystem integration, and business-technology alignment- which best predicts successful RAI practices. We discuss implications for management, policymakers, and AI governance standards.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

The role of AI assets and capabilities in shaping responsible AI deepening: a random forest machine learning view

  • 期刊论文

  • 发表日期：

    2025年10月21日

查看全文
  • Jacques Bughin

摘要

Responsible Artificial Intelligence (RAI) has emerged as a critical concern as organizations increasingly adopt powerful AI technologies without ensuring the development of organizational capabilities required to embed strong ethical practices into AI workflows at scale. In the absence of such capabilities, RAI may become symbolic rather than structural, with firms falling into ethics washing or governance gaps despite good intentions. This study investigates what mix AI assets and capabilities support strong RAI practices. Using longitudinal survey data from 193 AI-active large multinational firms between 2018 and 2024, we identify three distinct RAI adoption trajectories via Latent Class Analysis. We then use Random Forest machine learning to evaluate the best bundle of AI-related assets and capabilities in shaping these trajectories. If RAI practices adoption can be driven by a desire for regulatory compliance, our results clearly show that this is the orchestration of three capabilities, -AI talent managment, ecosystem integration, and business-technology alignment- which best predicts successful RAI practices. We discuss implications for management, policymakers, and AI governance standards.