Stakeholder Participation for Responsible AI Development: Disconnects Between Guidance and Current Practice

Responsible AI guidance promotes stakeholder involvement for reasons that differ from current industry practice: The study found that rAI documents recommend stakeholder involvement to achieve goals like redistributing power, improving understanding of social and technical issues, anticipating risks, and enhancing public oversight. However, practitioners (n=130 surveyed, n=10 interviewed) primarily engage stakeholders for commercial reasons such as customer value and compliance—suggesting a fundamental misalignment between rAI ideals and business reality.

Current stakeholder involvement practices in AI development are not adequately serving responsible AI goals: The research reveals that "established SHI practices are largely not contributing to rAI efforts," indicating that simply adopting stakeholder engagement methods from traditional software development is insufficient for addressing the unique ethical and societal challenges posed by AI systems.

Multiple barriers prevent companies from adopting more responsible AI-focused stakeholder practices: The study identifies "several factors currently discourage more rAI-aligned SHI practices," though specific barriers are not detailed in the abstract, suggesting that moving industry practice toward responsible AI requires understanding and addressing these structural or organizational obstacles.

There is a critical gap between what responsible AI experts recommend and what practitioners actually do: By analyzing 56 rAI guidance documents alongside real-world practitioner behavior, the study documents concrete evidence of a disconnect between aspirational rAI principles and commercial implementation, highlighting the need for interventions that bridge this gap.