This contribution explores how the integration of Artificial Intelligence (AI) into organizational practices can be effectively framed through a socio-technical perspective to comply with the requirements of Human-centered AI (HCAI). Instead of viewing AI merely as a technical tool, the analysis emphasizes the importance of embedding AI into communication, collaboration, and decision-making processes within organizations from a human-centered perspective. Ten case-based patterns illustrate how AI support of predictive maintenance can be organized to address quality assurance and continuous improvement and to provide different types of sup-port for HCAI. The analysis shows that AI adoption often requires and enables new forms of organizational learning, where specialists jointly interpret AI output, adapt workflows, and refine rules for system improve-ment. Different dimensions and levels of socio-technical integration of AI are considered to reflect the effort and benefits of keeping the organization in the loop.

Background: Active participation is a prerequisite for collaborative learning. In hybrid collaboration - where colocated and online participants collaborate on tasks - active participation of online participants is a challenge that may impact collaborative learning. Aims: The study aims to gain insights into whether a collaboration script and/or collaborative reflection promote active participation of online participants in hybrid collaborations. Sample: n = 88 university students with different academic backgrounds solved two collaborative tasks in groups of four or five. Methods: In a 2 x 2 between-subjects design, we conducted 20 hybrid collaboration sessions in which two to three co-located participants collaborated with two online participants. Depending on their experimental condition, groups 1) received no support, 2) used a collaboration script, 3) used a reflection scaffold, or 4) received both. Active participation was evaluated by analyzing the amount and duration of content-related contributions. We then compared the differences between co-located and online participants across conditions. Results: Our results suggest that, without instructional support, online participants participate significantly less than their co-located peers. Neither of our support measures affected this difference in participation in a statistically significant manner. However, our results provide some descriptive evidence that providing a script and/ or a reflection activity may help to narrow the participation gap. Conclusions: Our findings underscore the problem that online participants might participate less than co-located participants in hybrid learning settings. Collaboration scripts and collaborative reflections may be suitable means for reducing the participation gap. However, conducting rigorous studies that address these questions come with methodological challenges.

Active participation is essential for collaboration to unfold its potential for learning, regardless of whether collaborative learning takes place in co-located, online or hybrid settings. Unbalanced participation remains a challenge in hybrid learning, with online participants contributing less than their co-located peers. In previous work, we addressed this issue by designing a sociotechnical hybrid collaboration setting aimed at promoting equal participation during hybrid collaborations. Promoting social interactions through a static collaboration script and one-time awareness-based collaborative reflection reduced the participation gap between co-located and online participants, but the observed effects were not statistically significant. This limitation highlighted the need for more adaptive support mechanisms. In this context, generative AI has emerged as a promising approach, offering real-time, context aware interventions aiming to promote participation. Based on data of hybrid collaborations, literature and expert interviews on the integration of generative AI in a sociotechnical hybrid collaboration setting, we developed four clusters of design recommendations (1) participation feedback, (2) individual participation prompts, (3) group participation prompts and (4) procedural guidance prompts. Central to these is generative AI-based assistance during collaboration to promote balanced participation. Three key functions of this recommended assistance are (1) real-time participation visualization based on live transcript analysis, (2) targeted inclusion prompts to encourage under-participating participants, and (3) adaptive collaboration scripting that respond to group dynamics. The proposed design recommendations aim to guide implementation of hybrid collaboration settings that make collaborative learning in hybrid settings more fruitful.

Based on the literature and several practical examples of possible AI applica-tions, we outline the concept of intervenability. This new phenomenon is not covered by emergency shutdowns, workarounds, or the reconfiguration of automated systems. Intervenability instantiates the principles of control-lability, autonomy, oversight, and keeping humans in the loop in the context of AI. We provide a taxonomy that encompasses a range of possibilities for intervening activities and differentiates them regarding the mental effort of the users. This taxonomy extends the scope of interventions from real-time control of automated processes to AI-based discrete case-related decision-making. This is in accordance with human-centered AI, which seeks to combine human strengths with the usage of AI. We demonstrate how inter-venability can potentially contribute to the ongoing development of human capabilities on the one hand and to further technical improvement by recon-figuration of AI on the other. Exploring and collaboratively reflecting on the effects of interventions as an integral part of organizational practices is key to enabling this continuous improvement on both sides. Intervenability also provides further momentum for the design of an AI that can help realize in-terventions on its own and advance a smooth transition from intervention to reconfiguration of the AI.

A systematic literature review reveals that the role of AI as a sparring partner (SP) is often proposed but not systematically analyzed or defined. We propose the definition: An AI sparring partner (AI SP) interacts with users in a combination of a cooperative as well as a challenging, competitive mode, where AI is selected, customized or self-adapting to meet a level of skills that neither over- nor under-challenges the users. They can have the experience that they become better or are better than AI. AI as a SP can support creativity, extend viewpoints or foster learning and critical thinking either towards ideas and decisions or towards the AI itself. Sparring with AI can either be explicit, e.g. when AI simulates certain roles, or implicit, when AI is used to find out whether or how to perform better.

This work's aim is to take a look at scenarios and requirements for applying artificial intelligence as a company-side tool for occupational safety. The target domain to be focused is the internal safety or risk management respectively. The characteristic data is complex and heterogeneous, which can be hard to analyze regarding their inter-correlations and which in part consist of rare events, such as accidents. Artificial Intelligence and especially Large Language Models provide new approaches to this data problem and to also analyze and model hitherto unknown correlations and to identify corresponding risks. As a safety critical application, the support of risk assessments with artificial intelligence must be grounded in practically relevant scenarios and thoroughly elicited and refined requirements. In this work, we report on the findings from fifteen expert interviews for requirements elicitation and from three scenario workshops. We integrate requirements and scenarios to take a step towards building test scenarios, with which developers can check if they are on target with their systems.

Human-centered AI (HCAI) refers to guidelines or principles that aim on ethically oriented design of systems. We compare HCAI-guidelines with principles of socio-technical systems that emerged in the context of conventional information technology. The comparison leads to a revision of socio-technical heuristics by including aspects of AI-usage. The comparison reveals that continuous evolution is a basic characteristic of socio-technical systems, and that human oversight or interventions and the subsequent appropriation of AI-systems lead to continuous adaptation and re-design of the systems, if autonomy is collaboratively exercised. From a socio-technical point of view, the crucial requirement of transparency has not only to be fulfilled with technical features, but also by contributions of the whole system including human actors. It will be promising for using AI, if not only technical features, but organizational and social practices are socio-technically designed in a way that compensates shortcomings of AI.

In hybrid collaboration settings, co-located and remote learners simultaneously work together on shared material and joint tasks. Promoting equal participation of all group members in hybrid collaboration - especially the remote ones - is a particular, yet underresearched challenge. To address this research gap, in a previous study we identified 16 requirements for improving hybrid collaboration. We converted each requirement into multiple solution options that help to meet the respective requirement. As not all these options can be integrated into the design of a hybrid collaboration setting, we need to know how these options can be prioritized and which combination of them is most promising and should be further investigated. Thus, we conducted six focus group sessions with three collaborative learners (university students) each to gain further insights into the appropriate design. We used storyboards and scenarios with conceptual designs of the solution options for each requirement that needed final design decisions in order to discuss possible solutions. The results of our exploration helped to identify a hybrid collaboration setting that might be most promising for promoting equal participation: On the basic technical side, for example, each co-located participant should use their own camera in addition to one shared camera for the whole group. The identified setting also combines script-based support with a group awareness tool. Thus, it, provides guidance on how to approach group work to ensure equal participation, and additionally allows to reflect on whether contributions are being distributed equally.

In the transition from COVID-19 back to regular life, hybrid collaboration has gained increased attention. Currently, hybrid collaboration settings in higher education are not sufficient to support equal participation between co-located and remote students. In a first step, we conducted a literature review on the characteristics of co-located, remote, and hybrid collaboration settings involved in successful collaboration. The characteristics of these three types of collaboration were then used to identify design elements for hybrid collaboration settings that support equal participation in small-group work in higher education. In a second step, we designed an initial hybrid collaboration setting for higher education and implemented it in an exploratory study to identify additional features that support equal participation and to analyze collaboration between co-located and remote participants. From the results of this analysis, we identified 16 requirements to support equal participation in hybrid collaboration settings in higher education, which can serve as the basis for implementing and evaluating proposed solutions in a next step.

Was sind digitale Kompetenzen für Politikwissenschaftler*innen? Und wie können Lehrende zu deren Entwicklung beitragen? Der Band bietet Impulse für die Vermittlung digitaler Kompetenzen in Lehrveranstaltungen. Auf Basis der Unterscheidung von vier Kompetenzbereichen veranschaulichen konkrete Beispiele aus der Lehre, wie der doppelten Herausforderung der digitalen Transformation konstruktiv begegnet werden kann: Einerseits durch neue digitale Lehr-, Lern- und Arbeitsformen und andererseits durch die inhaltliche Auseinandersetzung mit den politischen sowie gesellschaftlichen Dimensionen des digitalen Wandels.

Occupational safety relies on systematic and reliable work place risk assessments. It is estimated, though that for high percentages of work places, no risk assessment is ever done. Since meanwhile technological progress and a far-spread lack of specialized personnel complicate matters, our idea is to use AI techniques to support work place risk assessment and thus to ultimately lower the requirements to get started. For such a tool to be accepted in the target context though, it needs to be usable, which requires a thoughtful interaction design. In this work we present the use case of work place risk assessment by providing an overview over the general process, underlying data, appropriate algorithms, and user requirements. We then proceed to match this use case against interaction design guidelines targeting human AI-interaction, especially those which understand human-AI interaction as interaction with a socio-technical system. By considering multiple guidelines, we further outline how the different guidelines come together in an actual use case.

This work demonstrates the variety of possible actions and reactions in the interplay of humans and AI when working on shared tasks. The coordination of human-AI task sharing has to take these varieties into account. In many instances the modes of reacting on each other are symmetrically distributed between humans and AI. Certain activities of coordinating the task sharing between human and AI are in a meta-relation to the interaction between humans and AI, some of these meta-activities may be reserved to humans.

This study emanates from work on human-centered AI and the claim of "keeping the organiza-tion in the loop". A previous study suggests a sys-tematic framework of organizational practices in the context of predictive maintenance, and identified four cycles: using AI, customizing AI, original task handling with support of AI, and dealing with con-textual changes. Since we assume that these findings can be generalized for other kinds of applications of Machine Learning (ML), we contrast the manage-ment activities that support the four cycles and their interplay with a widely different domain: the usage of AI for radiology. Our literature analysis reveals a series of overlaps with the existing framework, but also results in the need for extensions, such as holis-tic consideration of workflows or supervision and quality assurance.

In the context of maintaining technical equipment, AI is used to detect possible problems. Human specialists check whether a real problem is addressed, and, in this case, try to solve it. Furthermore, they go on trying to translate the problem notification into an improvement of other software components into which the AI system is embedded. Thus, every AI result is not only the cause of immediate action but is also a trigger within the process of continuous appropriation of the technical infrastructure that includes AI. The whole socio-technical system is a subject of AI-related improvement as a collaborative task that requires continuous advancement of human competences and skills. This has to be supported by a type of explainable AI by which the process of understanding the reasons driving AI output is not a task for a single end-user but rather the result of combining different specialists’ viewpoints and competences.

The human-centered AI approach posits a future in which the work done by humans and machines will become ever more interactive and integrated. This article takes human-centered AI one step further. It argues that the integration of human and machine intelligence is achievable only if human organizations—not just individual human workers—are kept “in the loop.” We support this argument with evidence of two case studies in the area of predictive maintenance, by which we show how organizational practices are needed and shape the use of AI/ML. Specifically, organizational processes and outputs such as decision-making workflows, etc. directly influence how AI/ML affects the workplace, and they are crucial for answering our first and second research questions, which address the pre-conditions for keeping humans in the loop and for supporting continuous and reliable functioning of AI-based socio-technical processes. From the empirical cases, we extrapolate a concept of “keeping the organization in the loop” that integrates four different kinds of loops: AI use, AI customization, AI-supported original tasks, and taking contextual changes into account. The analysis culminates in a systematic framework of keeping the organization in the loop look based on interacting organizational practices.

Modeling social-technical systems’ work processes as a basis for requirements engineering is a challenging issue. One of the most important pre-conditions for designing a socio-technical system is that system analysts know and understand how the system should support a company’s work processes, and what kind of intentions of the involved stakeholders influence each other and substantiate the requirements to be met. The goal of this paper is to provide a modelling method that helps the system analysts to develop this knowledge, and to evaluate the system’s features. Modelling notations for socio-technical systems –such as SeeMe– support the definition of activity sequences as well as the representation of contingency, explicit incompleteness, and flexibility. Other notations – such as i* – allow for representing goals and intentions of actors and the dependencies between these actors. The focus of this research is to support requirements engineering (RE) for socio-technical systems by merging these two modelling approaches. The result is a new modelling notation (SeeMe*) that extends the process-oriented view by modeling the agent-oriented paradigm of the i*-framework that covers agent properties such as in-tentionality, autonomy, sociality and boundaries. Furthermore, the integration of goals into process models supports socio-technical RE . With a case study in the context of a pharmacy, we demonstrate how SeeMe* enables analysts to systematically generate the specifications of solution-oriented requirements in early phases. We conclude that this approach can avoid determining too early whether the solution is based on socio-organizational measures or on technical components and infrastructures.

Abstract The construction industry is one of the sectors with the highest accident rates. To prevent accidents, construction workers receive occupational safety training and safety instructions. However, experience-based learning of dangerous situations is hardly possible or justifiable in reality. Virtual reality (VR) simulations can be a potential solution in this regard by allowing workers to experience dangerous situations in a very vivid but safe way without being exposed to real hazards. In this study, a VR simulation for construction safety training was developed and tested with trainees that learn the safe operation of hand-operated power tools. In this particular case study, the objective for the participants in the VR simulation was to successfully consider all safety aspects in the operation of an angle grinder. The usability, user experience and implicit learning were investigated during the study. Additionally, we conducted post-play interviews with participants. Results found learning effects of participants as well as a satisfying user experience and usability. The results also show that participants might learn content as presented, risking the learning of false information if the simulation does not cover relevant safety aspects.