The Application of AI Teachers in Facilitating Game-Based Literacy Learning

This chapter discusses the theoretical frameworks for artificial intelligence (AI) teachers and how AI teachers have been applied to facilitate game-based literacy learning in existing empirical studies. While the application of artificial intelligence (AI) in education is a relatively emerging research area, it has received increasing attention in the scientific community. In the future, AI teachers are likely to be able to serve as powerful supplementary tools in classroom teaching in support of human teachers. The main goal here is to provide the readers with new insights on promoting game-based literacy learning from the perspectives of AI teachers. To this end, the authors introduce the readers to the key concepts of AI teachers, the merits and demerits of AI teachers in education, scientific research on AI teachers in literacy learning, and some highlighted examples of AI teachers in literacy classrooms for practical concerns.

Implementing Learning Analytics: The journey to improve teaching and learning at five Australian universities

Universities are under heightened pressure to become more efficient using less resources., the quality of teaching and the student experience must not be sacrificed in pursuit of efficiency. One strategy is to use automation, smart technology to augment the work of human teachers. Not to replace the teacher but to make them better at what they do. Give them smart tools to do their jobs more effectively. Learning Analytics is one such tool that has the potential to leverage teaching capability. This paper examines the learning analytics implementations at five diverse Australian universities (regional and metropolitan) with varying degrees of success reported. These implementations are evaluated using of DeLone and McLean’s (2003) information system success model. It will be seen that participants in this interpretivist case study regard learning analytics as having potential benefits but are not sure about how best to realise analytics systems with extensive usability research built-in and offering sophisticated functionality seem likely to emerge and take precedence over the trial and error approach. This study addresses an apparent gap in the research as limited studies exist targeting both learning analytics and information system success.

Learning reward functions from diverse sources of human feedback: Optimally integrating demonstrations and preferences

Reward functions are a common way to specify the objective of a robot. As designing reward functions can be extremely challenging, a more promising approach is to directly learn reward functions from human teachers. Importantly, data from human teachers can be collected either passively or actively in a variety of forms: passive data sources include demonstrations (e.g., kinesthetic guidance), whereas preferences (e.g., comparative rankings) are actively elicited. Prior research has independently applied reward learning to these different data sources. However, there exist many domains where multiple sources are complementary and expressive. Motivated by this general problem, we present a framework to integrate multiple sources of information, which are either passively or actively collected from human users. In particular, we present an algorithm that first utilizes user demonstrations to initialize a belief about the reward function, and then actively probes the user with preference queries to zero-in on their true reward. This algorithm not only enables us combine multiple data sources, but it also informs the robot when it should leverage each type of information. Further, our approach accounts for the human’s ability to provide data: yielding user-friendly preference queries which are also theoretically optimal. Our extensive simulated experiments and user studies on a Fetch mobile manipulator demonstrate the superiority and the usability of our integrated framework.

TANGO: Commonsense Generalization in Predicting Tool Interactions for Mobile Manipulators

Robots assisting us in factories or homes must learn to make use of objects as tools to perform tasks, e.g., a tray for carrying objects. We consider the problem of learning commonsense knowledge of when a tool may be useful and how its use may be composed with other tools to accomplish a high-level task instructed by a human. We introduce TANGO, a novel neural model for predicting task-specific tool interactions. TANGO is trained using demonstrations obtained from human teachers instructing a virtual robot in a physics simulator. TANGO encodes the world state consisting of objects and symbolic relationships between them using a graph neural network. The model learns to attend over the scene using knowledge of the goal and the action history, finally decoding the symbolic action to execute. Crucially, we address generalization to unseen environments where some known tools are missing, but alternative unseen tools are present. We show that by augmenting the representation of the environment with pre-trained embeddings derived from a knowledge-base, the model can generalize effectively to novel environments. Experimental results show a 60.5-78.9% improvement over the baseline in predicting successful symbolic plans in unseen settings for a simulated mobile manipulator.

School-wide positive behaviour support as evidence-making interventions

Global implementation of School-Wide Positive Behaviour Support (SWPBS) has grown considerably over the last forty years. SWPBS seeks to provide a multi-tiered approach to strategically frame integrated actions responding to matters of wellbeing, discipline and punishment in schools using Evidence-Based Interventions (EBIs). In this respect, EBIs rely on what are presumed to be value-free, reliably generated data which direct the selection, implementation and monitoring of SWPBS in schools. The paper begins by exploring how SWPBS is understood via implementation science. Following an outline of standard SWPBS EBI practice, discussion turns to instead consider this work as Evidence-Making Interventions (EMIs). To do so, first we outline how our before-the-fact anticipations influence relationships and disciplinary actions as they are realised in schools through EBIs. Our focus then turns to explain how SWPBS interventions, using multi-tiered systems of support and technology as examples, might be understood and enacted differently if engaged as EMIs. In conclusion, school-based relationships are subsequently reconsidered as a confluence of human (teachers and students) and non-human (data and policy) liaisons always and already subject to each other’s next move.

The use of AI in education: Practicalities and ethical considerations

There is a wide diversity of views on the potential for artificial intelligence (AI), ranging from overenthusiastic pronouncements about how it is imminently going to transform our lives to alarmist predictions about how it is going to cause everything from mass unemployment to the destruction of life as we know it. In this article, I look at the practicalities of AI in education and at the attendant ethical issues it raises. My key conclusion is that AI in the near- to medium-term future has the potential to enrich student learning and complement the work of (human) teachers without dispensing with them. In addition, AI should increasingly enable such traditional divides as ‘school versus home’ to be straddled with regard to learning. AI offers the hope of increasing personalization in education, but it is accompanied by risks of learning becoming less social. There is much that we can learn from previous introductions of new technologies in school to help maximize the likelihood that AI can help students both to flourish and to learn powerful knowledge. Looking further ahead, AI has the potential to be transformative in education, and it may be that such benefits will first be seen for students with special educational needs. This is to be welcomed.