Andrej Karpathy, co-founder of OpenAI, pointed out that current AI agents remain far from achieving "full functionality." He highlighted key limitations, including inadequate intelligence and constrained multimodal capabilities, which significantly reduce their practical effectiveness. Karpathy projected that it could take approximately a decade to comprehensively resolve these challenges. While some investors have labeled 2025 as the "Year of the Agent," the reality is that AI agent development continues to grapple with substantial obstacles.

Quintin Au, Growth Lead at Scale AI, also voiced skepticism about the current state of AI agents. He emphasized that error rates tend to escalate as task complexity rises. For instance, the likelihood of successfully completing five consecutive operations without any mistakes stands at just around 32%, underscoring the pressing need for continued innovation and refinement in this field.