HAVA: Hybrid Approach to Value-Alignment through Reward Weighing for Reinforcement Learning
This addresses the challenge of creating AI agents that adhere to societal values, which is crucial for safe and trustworthy deployment, though it appears incremental in combining existing norm representations.
The paper tackles the problem of value-alignment in reinforcement learning by integrating both explicit (e.g., legal/safety) and implicit (e.g., social) norms into a single algorithm, using a reputation-based reward weighing method. Experiments on a traffic problem show the method successfully finds value-aligned policies and outperforms using either norm type alone.
Our society is governed by a set of norms which together bring about the values we cherish such as safety, fairness or trustworthiness. The goal of value-alignment is to create agents that not only do their tasks but through their behaviours also promote these values. Many of the norms are written as laws or rules (legal / safety norms) but even more remain unwritten (social norms). Furthermore, the techniques used to represent these norms also differ. Safety / legal norms are often represented explicitly, for example, in some logical language while social norms are typically learned and remain hidden in the parameter space of a neural network. There is a lack of approaches in the literature that could combine these various norm representations into a single algorithm. We propose a novel method that integrates these norms into the reinforcement learning process. Our method monitors the agent's compliance with the given norms and summarizes it in a quantity we call the agent's reputation. This quantity is used to weigh the received rewards to motivate the agent to become value-aligned. We carry out a series of experiments including a continuous state space traffic problem to demonstrate the importance of the written and unwritten norms and show how our method can find the value-aligned policies. Furthermore, we carry out ablations to demonstrate why it is better to combine these two groups of norms rather than using either separately.