Machines to Understand Human Goals

Human goals are an important kind of context, so the ability is broadly useful.

Building Machines That Better Understand Human Goals

MIT News   By Rachel Gordon

An algorithm that infers goals and plans, even when those plans might fail, has been developed by Massachusetts Institute of Technology (MIT) researchers. The team used MIT's Gen artificial intelligence (AI) programing platform to integrate symbolic artificial intelligence (AI) planning with Bayesian inference, into the Sequential Inverse Plan Search (SIPS) algorithm. SIPS produced a model that outperformed the baseline Bayesian Inverse Reinforcement Learning model up to 150-fold, and the algorithm inferred goals with 75% accuracy, MIT's Tan Zhi-Xuan said, "This ability to account for mistakes could be crucial for building machines that robustly infer and act in our interests." ... '