This paper develops an evaluation framework for emerging generative artificial intelligence (GenAI) tutoring tools by synthesizing research from self-regulated learning (SRL), self-determination theory (SDT), and instructional design principles. The framework addresses how GenAI tools can support independent learning. This is accomplished through four key dimensions: SRL phases (forethought, performance, self-reflection), SRL developmental stages (observation, emulation, self-control, self-regulation), SDT principles (autonomy, competence, relatedness), and First Principles of Instruction. The framework uses an Innovation Configuration Map to operationalize the dimensions. It examines what learners are trying to accomplish, how the AI responds, the kind of background information it uses, and how the interface is designed. The map is tested by evaluating ChatGPT’s Study and Learn feature. The results were mixed. Some aspects performed well, but there were clear gaps in the current generic GenAI implementation. Students need autonomy to learn independently, but they also need scaffolding to stay on track. Finding that balance is hard. The framework gives practitioners a way to choose and implement GenAI tools more thoughtfully.