Understanding and attributing mental states, known as Theory of Mind (ToM), emerges as a fundamental capability for human social reasoning. While Large Language Models (LLMs) appear to possess certain ToM abilities, the mechanisms underlying these capabilities remain elusive. In this study, we discover that it is possible to linearly decode the belief status from the perspectives of various agents through neural activations of language models, indicating the existence of internal representations of self and others' beliefs. By manipulating these representations, we observe dramatic changes in the models' ToM performance, underscoring their pivotal role in the social reasoning process. Additionally, our findings extend to diverse social reasoning tasks that involve different causal inference patterns, suggesting the potential generalizability of these representations.

Game developers strive to have engaging believable characters in their work. One of the elements that has been pointed out as contributing to believability is social behavior. A category of social behavior is interpersonal conflict. In our current research we compare two AI approaches to model NPC conflict resolution strategies: one using the reactive planning language ABL and another using the AI framework FAtiMA. We identify the following metrics to evaluate social behavior modeling: mapping theory, emotion, model checking, variability, policy change. In our analysis we found it was easier to map conflict concepts in ABL and the model checking process was faster. FAtiMA had better support for emotion and other emergent attributes.

Remember the recent Elon Musk and Mark Zuckerberg clash on the future of Artificial Intelligence? So, my colleague and I were discussing the topic and after a while she said she doesn't understand machine learning & Artificial Intelligence fully. Are you one of those, who understand the basics of AI, the robots and more; yet when it comes to deep and in depth technical understanding, it suddenly becomes confusing? If yes, worry not for you have landed at the right place. We'll try to understand Machine Learning like a beginner.