Goals as Reward-Producing Programs
This addresses the gap in computational models for capturing everyday human goals, which is incremental by building on existing work on goals and program synthesis.
The authors tackled the problem of modeling the richness of human-generated goals by collecting a dataset of playful goals as scorable games, representing them as reward-producing programs, and generating novel goals through program synthesis. They found that model-generated goals were indistinguishable from human-created ones, with internal fitness scores predicting games rated as more fun and human-like.
People are remarkably capable of generating their own goals, beginning with child's play and continuing into adulthood. Despite considerable empirical and computational work on goals and goal-oriented behavior, models are still far from capturing the richness of everyday human goals. Here, we bridge this gap by collecting a dataset of human-generated playful goals (in the form of scorable, single-player games), modeling them as reward-producing programs, and generating novel human-like goals through program synthesis. Reward-producing programs capture the rich semantics of goals through symbolic operations that compose, add temporal constraints, and allow for program execution on behavioral traces to evaluate progress. To build a generative model of goals, we learn a fitness function over the infinite set of possible goal programs and sample novel goals with a quality-diversity algorithm. Human evaluators found that model-generated goals, when sampled from partitions of program space occupied by human examples, were indistinguishable from human-created games. We also discovered that our model's internal fitness scores predict games that are evaluated as more fun to play and more human-like.