News
Companies Want Humans to Oversee AI, but Most Can't Monitor It Effectively
Companies are keeping humans involved in AI decision-making, but many lack effective systems to monitor AI performance, according to a new global study by Tata Consultancy Services (TCS).
The TCS Enterprise AI Study surveyed 1,333 senior business and technology leaders across 25 countries and 12 industries.
Researchers found that a startling 9 percent of people questioned rated their real-time AI performance monitoring as highly and consistently effective.
By comparison, 32 percent gave the same rating to their ability to manage exceptions, escalations and human intervention.
Human Oversight Remains Central to AI Decisions
According to the research, approximately two-thirds of respondents said their organizations either do not allow AI to execute decisions without human approval or permit autonomous decisions only in a few specific use cases.
The findings suggest that while businesses are increasingly using AI to support decision-making, they remain cautious about allowing systems to act independently.
However, the study identified a gap between companies' ability to intervene and their ability to detect when intervention is necessary.
Security, risk and governance considerations also ranked as the leading factors influencing where and how organizations develop and deploy AI, ahead of cost, speed and ease of deployment.
AI Governance Remains Fragmented
The research team found that 42 percent of organizations either manage AI governance separately across platforms and teams or rely primarily on individual vendors' basic governance capabilities.
Convert to PDF