As AI voice agents become common in high-volume contact centers, firms often split incoming calls: low-intent leads go to the AI, while high-intent leads are reserved for human workers. The logic seems straightforward—let automation handle the easy cases and give people the promising ones. But a new arXiv paper warns that this arrangement may not work as intended.

Drawing on sequential search theory, the authors posit that AI-induced task allocation changes how human service workers perceive their own role. When workers are handed only high-intent leads, they may see those opportunities as more valuable—and, paradoxically, respond in ways that undermine the firm's goals. The paper's title captures the central puzzle: better opportunities can backfire.

The study is theoretical, and the abstract is brief, so the exact mechanism is not fully detailed here. Still, it highlights a practical risk for firms deploying AI alongside human agents: optimizing call routing by lead quality alone may ignore how workers react to the new division of labor. The authors suggest that managers should consider these behavioral responses when designing parallel AI-human systems.