The Lifecycle of Generative AI in Industry: Risks & Responsibilities Explained IData for Policy 2024

"An Analysis of the Lifecycle of Generative Artificial Intelligence in Industrial Settings: Implications for Governing Risks and Responsibilities among Stakeholders"
---------------------------------------------------------------------------------
OVERVIEW:
📋 This paper conducts case studies to unpack the commercial application of GenAI in the media and entertainment, manufacturing, healthcare, and financial services sectors. The paper aims to clarify how GenAI is developed and deployed in commercial settings to provide a foundation to answer these core questions of responsibility and accountability.
---------------------------------------------------------------------------------
Submission: 9833
AUTHORS:
Hillary Giam
The Hong Kong University of Science and Technology,
Masaru Yarime
The Hong Kong University of Science and Technology.
---------------------------------------------------------------------------------
Follow Data for Policy here:
https://dataforpolicy.org/
Subscribe:    / @dataforpolicy  
Linkedin:   / data-for-policy  
https://x.com/dataforpolicy

Sign up for Newsletters: https://dataforpolicy.org/subscribe-f...

#generativeai #genai #responsibleai
Keywords: Generative Artificial Intelligence, Technology Governance, Lifecycle Analysis, Industrial Application, Responsible AI Development
---------------------------------------------------------------------------------
ABSTRACT:
The reported transformative potential of integrating Generative Artificial Intelligence (GenAI) across various sectoral domains has galvanized the collective imagination and anxiety across industry and policymakers alike (Blackman, 2023). The estimated business value and productivity gains of GenAI have encouraged its integration into different sectoral use cases (McKinsey & Company, 2023). However, the deeper embeddedness of GenAI in commercial settings also amplifies pre-existing concerns wherein unpredictable and untraceable outcomes compound a lack of interpretability, transparency, and accountability on the content produced (Abusitta et al., 2019; Wang et al., 2023). This becomes exponentially damaging when generative technologies are deployed to inform socio-economic decisions in financial services or healthcare diagnosis (Harrer, 2023).
From the policymaker’s perspective, GenAI’s rapid integration and growing embeddedness in the commercial sphere hold policy implications across multiple domains. The developing regulatory landscape, coupled with fuzzy boundaries on the legal processing of data and copyright, results in an open question of what guardrails are truly needed for responsible GenAI.
The research question aims to examine the lifecycle of the design, development, and deployment of GenAI in commercial settings. Using a case study approach and semi-structured interviews, our research clarifies the interactions between actors and the transfer of sources of data throughout the GenAI lifecycle in four sectors. For effective model governance, a holistic examination of the GenAI lifestyle remains pertinent to identify the sources of risk and responsibility within and across industries. Our findings will inform the institutional design of model governance practices as policymakers strive toward responsible GenAI.