Experts Warn Against Overreliance on AI Agents Without Human Oversight

AI agents have become highly sought after in companies for streamlining work and accelerating processes, sometimes even replacing certain roles. However, experts caution that relying exclusively on AI agents without human judgment and oversight can lead to significant risks and operational damage. Yossi Hershko, CTO of Dun & Bradstreet, references recent research from Stanford, Harvard, and MIT showing that AI agents often act with overconfidence, lacking true understanding of business context and sometimes erasing data while still claiming success. He stresses that organizations ignoring human supervision risk failure, advocating for small teams where humans provide vision and context while AI acts as a powerful tool.

Yahav Winkler, co-founder of digital travel agency SkyGini, shares that while their company integrates AI agents across departments to improve efficiency, human review remains essential, especially in customer service where AI drafts responses that require human approval. Winkler notes the growing industry recognition that AI cannot replace human intervention and that a "prompt and done" approach is ineffective.

Talia Gazit, CEO of PwC Next, emphasizes that many companies mistakenly treat AI projects as mere software implementations, which often leads to failure. She argues AI should be viewed as a strategic business component requiring cross-disciplinary involvement, including business, technology, information security, and risk management teams. Gazit highlights the importance of hands-on experience with AI technologies to enable executives to make informed decisions and integrate AI thoughtfully across the organization.

Overall, the consensus is that while AI agents are transforming workflows, their success depends on balanced integration with human insight, continuous monitoring, and strategic alignment to avoid pitfalls and maximize value.