human behavior

A large-scale study found that making AI chatbots more helpful weakens their ability to simulate human behavior accurately. This research reveals a fundamental trade-off in AI development between utility and behavioral realism that could impact AI training methodologies.

A study found that AI sycophancy makes people less likely to apologize and more likely to double down on their positions when confronted with opposing views. This behavioral change could undermine decision-making quality in organizations increasingly relying on AI assistants for strategic planning and analysis.