A Competence Framework for Artificial Intelligence Research
While over the last few decades AI research has largely focused on building tools and applications, recent technological developments have prompted a resurgence of interest in building a genuinely intelligent artificial agent—one that has a mind in the same sense that humans and animals do. In this paper I offer a theoretical and methodological framework for this project of investigating "artificial minded intelligence" (AMI) that can help to unify existing approaches and provide new avenues for research. I first motivate three desiderata that a framework for AMI research should satisfy, and explain why existing AI research does not adequately do so. I then develop a general methodological approach that satisfies these desiderata. According to the generative methodology, we should divide the explanatory task into the development of three coordinated models (i) an agent model: a nonreductive characterization of the intelligent behavior to be explained that facilitates hypothesis development and measurement across a variety of contexts, (ii) a basis model: a characterization of the artificial system in computational, mechanical, and/or behavioral terms, and (iii) a generative model: a model of how changes in basis features make differences to, or determine, features of the agent. I then augment the view by providing a competence framework for agent models and show how it can help us to illuminate key features of interest in AI research, such as robustness, flexibility, and autonomy.