Perceptions of Agentic AI in Organizations: Implications for Responsible AI and ROI

Knowledge gaps are a critical barrier: AI professionals report significant gaps in understanding how to implement responsible AI frameworks for agentic systems, which are more autonomous and complex than traditional AI. This lack of knowledge directly undermines organizations' ability to deploy these systems responsibly and achieve return on investment.

Stakeholder engagement is being neglected: The research found that organizations place limited emphasis on engaging stakeholders—including employees, customers, and affected communities—in responsible AI implementation. This narrow approach risks missing important perspectives needed to identify and mitigate potential harms from autonomous AI systems.

Over-emphasis on control creates implementation problems: Organizations tend to focus heavily on controlling agentic AI systems rather than balancing control with other responsible AI dimensions. This control-centric approach actually hinders effective adaptation and implementation, suggesting that a more holistic approach to responsible AI governance is needed.

Interconnected dimensions require integrated solutions: Responsible AI involves multiple interconnected dimensions that cannot be addressed in isolation. The paper's findings suggest that organizations struggling to implement these frameworks often treat them as separate concerns rather than as an integrated system, leading to incomplete and ineffective responsible AI practices.

Complexity of agentic AI demands new organizational approaches: The novelty and inherent complexity of increasingly autonomous AI systems are outpacing current organizational adaptation capabilities, creating a gap between the sophistication of the technology and the maturity of governance frameworks designed to manage it responsibly.