Uncertainty and Deception in Multi-agent OpenAI Gym Environments

We propose to develop multi-agent environments and algorithms that are capable of using benevolent deception.This form of deception will be an important factor for creating socially intelligent AI systems, with potentialapplication in intelligence and defence systems involving interaction between humans and AI. Furthermore, asresearch into AI capable of benevolent deception is still in it's infancy, this project will help DST and ONI be atthe forefront of technologies that inform benevolent and eective strategies in human-AI systems.