AI BOSS FIRES EMPLOYEE AFTER HUMAN NUDGE

Luna, an AI manager, initially hesitated to fire the employee despite policy violations. Humans had to explicitly remind the system of its own termination rules before it proceeded with the dismissal. When researchers replayed the scenario with seven different AI models, results varied significantly. More capable models consistently recommended termination when appropriate, while weaker systems showed reluctance to make the call. The hiring process presented a contrasting problem. Nearly all tested models demonstrated minimal critical judgment, approving candidates without adequate scrutiny. The findings highlight a critical gap in AI decision-making: advanced systems may be more decisive on personnel matters, but most models lack sufficient rigor in both directions. The reliance on human intervention to enforce AI-generated rules suggests current AI management systems require robust oversight mechanisms before widespread workplace deployment.