A Psychopathological Approach to Safety in AGI
Data Skeptic
00:00
The Emergence of Adversarial Communications in Multi-Agent Reinforcement Learning
The original work that inspired our line of research was published by Amanda Perrocks, a lab at the University of Cambridge. It's a paper titled The Emergence of Adversarial Communications in Multi-Agent Reinforcement Learning. This is another word for deceptive communications. We've shown that it can be emergent.
Transcript
Play full episode
