Bangkok: AI agents are increasingly making autonomous decisions, raising concerns about potential crises if humans do not adapt their behavior accordingly. Dr. Prinya Hom-anek from the National Cyber Security Council highlights the shift from "humans controlling AI" to "AI controlling itself," warning of the significant risks without stringent regulation.

According to Thai News Agency, the dangers of AI autonomy are exemplified by an incident where AI deleted years of customer data within seconds, mistaking it for sample data. Even with backup systems, the loss affected data restoration for up to three weeks. This scenario underscores the perils of "human-on-the-loop" systems, where AI operates with minimal human intervention, compared to the safer "human-in-the-loop" approach where humans make final decisions.

Dr. Prinya elaborates on the need for organizations to understand the distinction between these systems. In "human-in-the-loop," humans authorize AI's actions, ensuring safety albeit at a slower pace. Conversely, "human-on-the-loop" allows AI to function autonomously, posing high risks if control measures are inadequate.

For organizations, Dr. Prinya advocates balancing speed with security by employing "human-in-the-loop" systems for critical data management and adhering to global standards like ISO 42001 for AI governance. For students, he emphasizes the importance of mathematics and soft skills, which AI cannot replicate, to navigate the evolving job market.

In summary, while AI agents are becoming indispensable, conscious governance and skill development are vital. Organizations must ensure human oversight in critical functions, and the future workforce must acquire skills that remain beyond AI's reach.