Artificial intelligence is currently one of the most discussed topics worldwide. It is therefore not surprising that the topic is also playing a role in education. This ranges from the development of special courses to claims that artificial intelligence should be integrated into national curricula. However, transformation in this field starts with teacher education. It is still an open question how bringing AI to schools can be realized on all required levels. This article discusses a framework and conditions taking all levels and stakeholders into account, starting at the national curriculum level. The complexity of processes are shown, the key levels and stakeholders are identified. The methodological approach and the results of an analysis of descriptions of the introduction of computer science as a compulsory subject from lower secondary level onwards from around 20 federal and central states are then presented. One result is the identification of six central measures. These measures are described and categorised in the CAPE framework. The descriptions were analysed with regard to the effect relationships between the six measures. Based on this analysis, a heuristic model of the work and development of curricula was developed. As a result, it is shown that at least four of the six measures must interlock in a meaningful way for an effective change in a national curriculum - independent of the topic. AI is then used as one exemplary content. The results of selected studies, including possible effect relationships, are presented and discussed as examples for each measure. Finally, this results in a guideline including: measures to be taken, the relationship between the effects of the measures and challenges to be considered.

To start the development of an optimal digital learning environment for language learning we have started with the first question: What current language learning software is available to users? To answer this question an analysis of current computer Assisted Language Learning (CALL) software is necessary. This paper expounds on the systematic analysis of 69 current language software. Based on this structured analysis, we have developed a first framework of requirements for an digital language learning environment. For this purpose, control, user input, software feedback and theoretical educational frameworks were analysed within this investigation. The analysis demonstrates a lack of constructivist frameworks and a higher prevalence of behaviourist educational frameworks in current language software.

Through the integration of the linguistic and computer science perspective of corrective feedback, this paper seeks to examine the forms of feedback present in current CALL software. This paper also expounds on the similarities and differences in the identification of feedback from these perspectives.

The trend Gamification occurred several years ago in the context of digital teaching and training aka e-learning. Looking closer at the development of game-based learning and gamification in computer science, it has to be admitted, that game-based learning seems to have ended in an impasse point where the instructionally smooth integration of learning and gaming has not really been realizable, yet -- at least not without enormous effort by the learning and gaming content author. The situation in gamification in digital teaching and training systems comes in a similar shape. A structured analysis of what gamification is can help to develop a systematic approach, which then can support a structured and methodologically smooth design of gamification for digital teaching and training.

This paper develops a structured approach towards learning. The current state of the art in using this term is ambiguous. It is unclear whether or not something can be perceived to be blended or not. In the current way of using this term, the term itself it obsolete. The missing point is a clear definition of what exactly the blend is. In many cases, learning is used to describe scenarios where traditional classroom education is enriched with technology, aka. Elearning. However, this is not enough to grasp the deeper meaning of learning. Thus, in the paper we will develop a general model of learning, which is based on a dialectic approach, and which shall help to structure the field. Examples support our approach.

In this paper we describe an approach to model activities for educational role playing games based on a cognitive task analysis. After developing an architecture for educational role playing games, including patterns of communication, we analyzed several games to find out, why it is so difficult to seamlessly integrate gaming and learning. Our analysis of activities in commercial RPGs showed us that activities like crafting are underdeveloped and play a minor role in game design (in contrast for example to combat). We have been able to show that crafting can be mapped to inquiry learning, so this might be a point of integrating learning and gaming. On this basis of results, we used cognitive task analysis to analyze the tasks to perform in classical crafting activities. We describe our approach with one example, the preparation of a recipe, in this paper.

In the context of an EU project we discuss design decisions regarding virtual experiments contrasting discovery learning and experiential learning approaches.

The paper proposes a component-based approach for the development of educational business simulation games. We identify common features of educational business simulations and construction and management games, and merge these in a reusable and extensible software architecture.

Design patterns are reusable solutions to recurrent problems. Generally described in the context of object-oriented programming, only a few contributions can be found in the domain of eLearning in general or Intelligent Tutoring Systems (ITSs) in particular. Based on several approaches to describe the systems' underlying architecture into pattern collections, we have developed a component-based framework. This framework provides a foundation for the development of flexible and reusable ITS. It also provides a basis for development of other eLearning systems. Still, the complexity of the inter-modular communication requires experience and self-discipline from the programmer. Thus we investigated the framework for additional patterns. We extracted useful communication patterns and integrated them into an editor. Patterns and editor are presented in this paper as they already proved themselves very useful in regard to programming efficiency and consistency.

Boundary Objects (BOs) are means of communication support for teams. In eLearning projects, teams usually consist of people with diverse professional backgrounds and different experiences. Due to this fact, team members have different perspectives on the eLearning product or service to be developed - eLearning system development thus becomes a complex and demanding task. Using specific BOs in a purposeful way can help people to communicate and support interaction between team members. Amongst others, process models, scripts, patterns and architecture models have been investigated as BOs for e-Learning system development.

Process models have been described by researchers starting in the mid 1990. Current state of the art in technology enhanced learning (TEL) systems development in combination with observations based on using of standards, existing process models, and on supervising team work in a series of diverse TEL systems, have led to the insight, that an extended approach is required. With a strong background in systems engineering, this led to the development of ROME (The Rostock model for systematically developing e-Learning products and services). From the perspective of ROME, the process model is an ideal type description of an interdisciplinary development process.

Intelligent tutoring systems can look back on a long tradition. Nonetheless, formal models are rare in this scientific domain. Thus, even if there is common agreement about the ITS architecture, the interpretation of the role and functionality of the components is quite heterogeneous. This leads to problems regarding reusability and comparability of components, and makes communication about components difficult. Learning and training based on a narrative story line makes special demands on ITS. This case-based training can be formally described by a tutoring process model. The tutoring process model is a domain independent model. It facilitates communication, comparability, and it supports re-usability. As a side effect, the formal tutoring process model leads to a new perspective on the ITS architecture. As a basis for implementation, the formal tutoring process model proved its usability and viability in the intelligent tutoring system Docs 'n Drugs $the virtual hospital.

A multitude of adaptive computer-based training systems have been implemented in the last years. Whereas several aspects of these systems necessarily are dependent on the application domain, including the learning style and the system type, some other parts of the systems are very similar: usually they require an execution component, which steers and controls the interaction between the learner and the underlying data of the application domain, e.g. databases. For reusing purposes we have developed a highly flexible and adaptable domain independent execution system that allows the realization of different learning strategies in a web-based scenario. In addition to steering the interaction between learner and data, the execution system is enriched with model checking facilities. They provide a verification mechanism for learning material and thus can be used to support the authors. The described system is integrated into the web- and case-based project "Docs 'N Drugs".

The authors discuss the structure of training cases in Web based and case-oriented training systems. We combine the benefits of systems that support a structural representation of knowledge underlying a training case and those that allow a fine granularity of didactical aspects. Our results emerged from the development of a training system in medicine, "Docs 'n Drugs-The Virtual Polyclinic" (T. Illmann et al., 2000). The abstraction from the proposed training case structure offers a good basis for general case oriented training systems, an easy exchange of tutoring data, and the Web based indexing of learning material.