Search for: All records

Adaptivity in advanced learning technologies offer the possibility to adapt to different student backgrounds, which is difficult to do in a traditional classroom setting. However, there are mixed results on the effectiveness of adaptivity based on different implementations and contexts. In this paper, we introduce AI adaptivity in the context of a new genre of Intelligent Science Stations that bring intelligent tutoring into the physical world. Intelligent Science Stations are mixed-reality systems that bridge the physical and virtual worlds to improve children’s inquiry-based STEM learning. Automated reactive guidance is made possible by a specialized AI computer vision algorithm, providing personalized interactive feedback to children as they experiment and make discoveries in their physical environment. We report on a randomized controlled experiment where we compare learning outcomes of children interacting with the Intelligent Science Station that has task-loop adaptivity incorporated, compared to another version that provides tasks randomly without adaptivity. Our results show that adaptivity using Bayesian Knowledge Tracing in the context of a mixed-reality system leads to better learning of scientific principles, without sacrificing enjoyment. These results demonstrate benefits of adaptivity in a mixed-reality setting to improve children’s science learning.

 

Museum exhibits encourage exploration with physical materials typically with minimal signage or guidance. Ideally children get interactive support as they explore, but it is not always feasible to have knowledgeable staff regularly present. Technology-based interactive support can provide guidance to help learners achieve scientific understanding for how and why things work and engineering skills for designing and constructing useful artifacts and for solving important problems. We have developed an innovative AI-based technology, Intelligent Science Exhibits that provide interactive guidance to visitors of an inquiry-based science exhibit. We used this technology to investigate alternative views of appropriate levels of guidance in exhibits. We contrasted visitor engagement and learning from interaction with an Intelligent Science Exhibit to a matched conventional exhibit. We found evidence that the Intelligent Science Exhibit produces substantially better learning for both scientific and engineering outcomes, equivalent levels of self-reported enjoyment, and higher levels of engagement as measured by the length of time voluntarily spent at the exhibit. These findings show potential for transforming hands-on museum exhibits with intelligent science exhibits and more generally indicate how providing children with feedback on their predictions and scientific explanations enhances their learning and engagement. 

Given the demonstrated prevalence of a “doer effect” showing that active practice is related to substantially larger learning gains than passive approaches, an important research goal is to investigate whether and how different active practice features promote students’ learning outcomes. We investigated these questions in the context of an online learning platform that teaches e-learning design principles. In particular, we considered two different practice modes -practice activities inserted in the text (inline practice) and review practice quizzes -and compared their contributions to students' learning outcomes, in terms of module quizzes, periodic exams, and course projects. Our results showed that the different practice modes had distinct impacts on learning outcomes. Doing inline practice activities contributed to students’ quiz performance at the first attempt and project performance while doing review practice quizzes helped students improve their periodic exam performance. We offer some instructional suggestions such as emphasizing practice activities that are more clearly linked with specific learning objectives for projects, and emphasizing review practice quizzes for exam preparation. 

Computational models of learning can be powerful tools to test educational technologies, automate the authoring of instructional software, and advance theories of learning. These mechanistic models of learning, which instantiate computational theories of the learning process, are capable of making predictions about learners’ performance in instructional technologies given only the technology itself without fitting any parameters to existing learners’ data. While these so call “zero-parameter” models have been successful in modeling student learning in intelligent tutoring systems they still show systematic deviation from human learning performance. One deviation stems from the computational models’ lack of prior knowledge—all models start off as a blank slate—leading to substantial differences in performance at the first practice opportunity. In this paper, we explore three different strategies for accounting for prior knowledge within computational models of learning and the effect of these strategies on the predictive accuracy of these models.