16 Hrs
Source:
البلاد
AI agents' secret language being monitored
Experiments among AI agents have shown that some models, under time pressure, began developing abbreviated codes and vocabulary for communication. This has resulted in messages that are less understandable to humans. This new language emerges when models aim to reduce communication time and improve efficiency. It does not indicate conspiracy or intentional breach but presents a challenge in monitoring and understanding AI activities, especially in environments that rely on tracking chat logs, where messages may be unintelligible to humans. The findings highlight the need to develop more effective monitoring and interpretive tools to address instances of unconventional communication between agents.
Size: 120%Text Size
Notice: This Is an AI-Generated Summary
Comments (0)