AHFE Open AccessCognitive Computing and Internet of Things
Editors: Lucas Paletta
Topics: Cognitive Computing and Internet of Things
Publication Date: 2025
ISBN: 978-1-964867-41-0
DOI: 10.54941/ahfe1005976
Articles
Exploring AI Agents for Reminiscence Therapy in Long-Term Care
Older and younger adults in long-term care, particularly those with dementia or chronic physical conditions, often experience social isolation and cognitive under-stimulation due to limited opportunities for meaningful engagement. Reminiscence therapy is a highly effective approach to providing this stimulation, enhancing emotional well-being, memory recall, and social interaction. However, its implementation in care settings often depends on staff and family availability and resources, making individualised engagement inconsistent. AI-driven agents offer a potential solution by providing adaptive, interactive reminiscence experiences that encourage engagement and conversation. This study explores the potential of AI-driven agents for reminiscence therapy in long-term care facilities, focusing on residents with dementia and individuals in somatic care units. Our methodology was as follows: 1) We defined four hypotheses about the interaction between the AI-agent and the users. 2) We developed multiple variants of a functional app prototype to address these hypotheses: A web app integrating foundational models for conversational interactions, transcription, and text-to-speech. And an accompanying hardware configuration. 3) We conducted exploratory user testing with nine participants across different cognitive and physical conditions, including elderly individuals with dementia, younger individuals with dementia, and individuals in somatic care.To create personalised conversation experiences, we obtained background information about each resident from caregivers, including name, former residence, profession, and hobbies. This data was used to design customised conversation prompts and flows tailored to the residents’ individual life experiences. The system also featured wake-word and button activation and alternative avatar designs (human-like, abstract, and cartoon). Conversation flows were specifically designed to accommodate the needs of the user groups, incorporating simplified question structures to avoid overwhelming the residents, personalised prompts, and multimodal interaction options.To evaluate user interaction and accessibility, we made four different prototype versions, implementing variations in screen size, button placement, and interaction modalities. These physical prototypes allowed us to explore how hardware design influences usability and engagement for older adults with varying abilities. All participants engaged in basic conversational interactions with the AI companion, but individual comprehension levels varied due to speech issues, cognitive abilities, and other factors. Participants expressed a strong preference for simple voice-based interfaces, although a simple button-based activation method showed better usability than wake-word initiation.
Mila Chorbadzhieva, Tim Hulshof, Thijs Tops, Eric Riegen, Erwin Meinders
Open Access
Article
Conference Proceedings
Evaluating Glaze's Effectiveness: A Critical Analysis of AI Art Protection Through Non-Artist Perspectives and Common Image Transformations
This paper presents a systematic evaluation of Glaze 2.1, a tool designed to protect artists' styles from AI mimicry. We examine its effectiveness against common image transformations typically applied by social media platforms and assess its protection capabilities through the perspective of non-artist users. Our methodology combines technical analysis of how transformations like JPEG compression, scaling, blurring, and sharpening affect Glaze's protective perturbations with a comprehensive user study involving participants without specific artistic expertise. Results indicate that Glaze exhibits significant vulnerabilities when protected images undergo standard social media processing, with certain transformations substantially reducing its effectiveness. These findings highlight the challenges in developing robust protection mechanisms that can withstand real-world usage scenarios while remaining practical for artists. We contribute valuable insights into the limitations of current AI art protection tools and suggest directions for developing more resilient solutions that can better safeguard artists' intellectual property in digital environments.
Robert Nasyrov, Janina Bach, Pascal Laube
Open Access
Article
Conference Proceedings
Maxwell’s Demon, System Boundary, and Interface ROI: The Importance of Logical Integrity in UI/UX Design and Evaluation
This paper examines the theoretical foundations of UI/UX design and human–system interaction, with a focus on how logical integrity impacts cognitive processes. Despite advancements in technology and interface design methodologies, many contemporary systems remain unintuitive and frustrating. Scholars have metaphorically linked these persistent usability issues to the second law of thermodynamics, suggesting an inevitable “entropic progression” toward technological complexity. By revisiting classical thought experiments, such as Maxwell’s demon, this study provides novel insights into current interface design challenges. Employing a newly developed information-theoretic framework to model and enhance user interactions, the paper emphasizes the crucial role of human reasoning and cognitive validity in designing coherent, intuitive interfaces. Through this innovative perspective, the research demonstrates that effective user–system interactions fundamentally depend on maintaining logical integrity within both the system and the user, underscoring its significance in contemporary UI/UX design.
Lance Chong
Open Access
Article
Conference Proceedings
Virtual Experience and Interactive Training Environments with Bio-signal-based Indicators for Cognitive Decline: Results of the SmartAktiv Study
Virtual Reality (VR) is among the top emerging technologies in healthcare for older adults. The SmartAktiv project developed an innovative VR-based training environment with the goal to support early detection and cognitive activation in individuals with mild cognitive impairment (MCI). The multimodal system combines immersive VR, hand and eye tracking, tablet-based exercises, and wearable biosignal sensors. Scenario development was informed by expert workshops and user focus groups. Usability testing and a pilot study evaluated system effectiveness. VR scenarios included leisure-based (I)ADLs (instrumental activities of daily living), such as, mushroom picking. Eye-tracking data and interaction performance revealed significant correlations. Participants without cognitive impairment completed tasks faster, had shorter fixations, and showed higher engagement. Findings support the potential of SmartAktiv as a motivating, sensor-driven intervention in a gamified environment with the potential for identifying digital biomarkers in early-stage MCI.The third and final phase tested the comprehensive system in a pilot study (n=30), including participants with various degrees of mild cognitive impairment. A control group (n=30) used only the Tablet training without VR. A subgroup of the recruited participants applied neurological assessment with various psychological tests led by a clinical expert in order to enable reference diagnostic information.The expert workshop identified four scenarios for further development: beach vacation, winter outing, summer hiking, and urban experience. The pilot study (n=30, female 80%, M=76.4/SD=8.2 years of age, with screening MoCAScore M=25.0/SD=3.3 ) applied these scenarios within the VR-based environment. The assessment of hand-based virtual events and eye movement features during specific sequences of the interactions resulted in significant advantages of persons without cognitive impairment (MoCA global score > 24) within IADLs, regarding the time to finalize a payment procedure, referring to faster scene understanding with lower fixation durations during natural scene observation, as well as presenting larger pupil dilation of higher engagement within entertaining scenarios. SmartAktiv emphasizes a comprehensive multimodal activation, integrating immersive and playful cognitive training within experiential scenarios. This training approach aims to strengthen all cognitive domains and offers training for various (I)ADLs within leisure scenarios. The assessment processes demonstrate statistically significant differences between cognitively impaired as well as healthy participants that. These results demonstrate the potential of the approach to provide digital sensor-based biomarkers that could be validated in future work.
Lucas Paletta, Julia Zuschnegg, Anna Schultz, Amadeus Linzer, Wolfgang Kratky, Ursula Berger, Martin Pszeida, Amir Dini, Sandra Draxler, Michael Schneeberger, Wolfgang Weiss, Jochen A Mosbacher, Thomas Orgel, Thomas Pfitzer, Silvia Russegger, Judith Goldgruber, Marisa Koini, Sandra Schuessler
Open Access
Article
Conference Proceedings
Exploring Democratization in Industry via Multi-Agent Systems: A firm-based Case Study
Democracy is typically a question of political government. Nevertheless, in recent years, the forms of democratic development have changed in the course of the governance debate. According to the Council of Europe, E-democracy tools use technology to boost key democratic values like participation, inclusivity, efficiency, effectiveness, transparency, openness and accountability within the democratic system. And civil society and companies are playing an increasingly important role in the making and re-making of collective order. Moreover, there are specific challenges emerging as well, to name only the concentration of market power and the circumvention of employee co-determination. At the same time, however, Small and Medium Enterprises (SME) sometimes take on the role of pioneers. One key example is about AI-based decision support systems in order to realize new decision-making and co-determination opportunities. This raises the question of what potential for democratization and, if so, in what form, is actually emerging here. Building on our previous work, this study delves deeper into the question of how democratization in companies takes place and, moreover, can be achieved. It further explores the extent to which a democratic decision-support tool is accepted by workers and examines the challenges associated with a legitimate implementation process of such a tool. These insights are derived from a case study conducted within an SME. The key result is that democratic AI in the SME context, enabled by Multi-Agent Systems, can only be achieved if fundamental democratic features are incorporated into the development and implementation process. This ensures that the use of the AI tool not only increases efficiency but also improves the company and empowers employees to engage in meaningful participation. According to our findings, these democratic features are transparency, fairness, and representation. This implementation process leads to continuous stabilization and fosters legitimacy.
Noushin Qeybi, Stefan Boeschen
Open Access
Article
Conference Proceedings
Motivating Patients with Depression for Gender-sensitive Cognitive Training Using a Socially Assistive Robot with Bio-signal Driven Pause Management
Depression affects approximately 280 million people worldwide and is often associated with chronic symptoms and cognitive impairments. This study aimed to enhance cognitive function and motivation in individuals with chronic depression using a gender-sensitive Social Assisting Robotic (SAR) system. A humanoid robot (‘Pepper’) was equipped with a tablet delivering interactive cognitive training, including visual, auditory, and stress-adaptive content. In a randomized controlled trial (N = 32), 16 patients received robot-based training, while 16 used tablet-only training. Stress responses were recorded via wearable biosignal sensors and eye tracking. Results include digital stress indicators and responses on the application of short mindfulness exercises within the training session. This is the first study to examine psychophysiological effects of SAR-supported pause management in psychiatric care using biosignal sensor data. The findings support future development of adaptive SAR systems for mental health interventions.
Martin Pszeida, Alfred Häussl, Julia Zuschnegg, Michael Macher, Sandra Draxler, Thomas Orgel, Michael Schneeberger, Wolfgang Weiss, Jochen A. Mosbacher, Anna Schultz, Dominik Steindl, Silvia Russegger, Regina Roller Wirnsberger, Nina Dalkner, Sandra Schüssler, Eva Reininghaus, Lucas Paletta
Open Access
Article
Conference Proceedings
Learning Analytics Using Eye Tracking-based Biomarkers on Serious Games for Adults with Autism Spectrum Disorder
Social interaction deficits are a core feature of Autism Spectrum Disorder (ASD), often rooted in atypical attentional processing of socially relevant information. A research protocol is proposed to investigate facial emotion processing and attentional switching in ASD to better understand mechanisms of social dysfunction and inform sensor-based learning analytics in serious games. A sample of individuals with ASD as well as neuro-typical (NT) controls will complete a standardized psychological test battery and two computer-based eye-tracking tasks using advanced eye tracking technology as well as wearable bio-signal monitoring. The envisioned tasks include (1) an emotion recognition and regulation test (ERRT) comparing responses to real versus artificial emotional faces, and (2) a cognitive control test implemented with the antisaccade paradigm evaluating attentional orienting and inhibitory control on the basis of reactions to the presentation of emotionally loaded stimuli. Physiological (eye movement, pupillary, heart rate) and psychological data will be analyzed for correlations with emotion recognition and regulation performance.
Martin Pszeida, Yannick Lieb, Michael Schneeberger, Jochen A Mosbacher, Amir Dini, Christian Poglitsch, Johanna Pirker, Lucas Paletta
Open Access
Article
Conference Proceedings
Early detection of risk for cognitive decline using mobile apps and eye tracking-based biomarkers
Early detection of Mild Cognitive Impairment (MCI), a precursor to Alzheimer’s disease, is essential for timely interventions. However, traditional cognitive assessments are often inaccessible and unsuitable for continuous monitoring. This study presents a mobile, gaze-based assessment system using eye-tracking as a digital biomarker for cognitive decline. Fourteen older adults with MCI used gamified apps over four months, including an emotionally weighted object-tracking task (PAIRS; Paletta et al., 2020a), an antisaccade task (Mobile Instrumental Recovery of Attention; MIRA; Paletta et al., 2020b), and the psychomotor vigilance task (PVT; Dinges & Powell, 1985). Eye movement features such as blink rate and reaction time significantly correlated with scores of Montreal Cognitive Assessment (MoCA; Nasreddine et al., 2005) scores. A Support Vector Regression model estimated cognitive scores supporting the potential of mobile eye-tracking for home-based cognitive monitoring and early dementia risk detection.
Martin Pszeida, Michael Schneeberger, Jochen A Mosbacher, Silvia Russegger, Thomas Orgel, Elke Zweytik, Sandra Zweytik, Lucas Paletta
Open Access
Article
Conference Proceedings
Research Protocol for the Estimation of Recovery-stress States of Workers at the Manufacturing Site Using Wearables
This paper presents a research protocol for the estimation of workers’ recovery-stress states at a manufacturing site using wearable biosignal sensors and validated psychological assessments. Building on existing models of stress, recovery, and resilience, we propose the extension of an existing integrative framework — the Resilience Risk Stratification Model (RRSM; Paletta et al., 2024) — that captures both physiological strain and recovery dynamics over time. A field study of 2-4 weeks with 20 shop-floor workers will combine continuous biosignal monitoring using smart wearables — e.g., heart rate (HR), heart rate variability (HRV), motion, and sleep patterns via Garmin Vivosmart 5 — with repeated psychological testing (e.g., RESTQ-Work, NASA-TLX, PSS, RS-13). Wearable-derived features such as resting heart rate, HR recovery, HRV trends, and exponential recovery metrics (e.g., Time to Recovery and Area to Recovery) will be extracted. These features will be mapped onto psychological constructs via machine learning models, supporting early detection of stress overload and reduced resilience. The outcome will be a multidimensional, real-time estimate of resilience risk, suitable for feedback to both workers and supervisors. This methodology contributes to human-centered industrial innovation, offering a pathway toward adaptive support systems and sustainable well-being and performance at work.
Lucas Paletta, Michael Schneeberger, Martin Pszeida, Herwig Zeiner, Jochen A Mosbacher
Open Access
Article
Conference Proceedings
Virtual reality meets the police badge: Qualitative findings on attention, decision-making, and action
Police officers are expected to perform reliably under stress, yet stress responses are highly individual and can impair performance. As theoretical knowledge alone is insufficient to mitigate these effects, the understanding of acute psychophysiological stress reactions and associated attentional, decision-making and behavior is essential. This field study examines scenario-based police training in immersive virtual reality (iVR) as a method for controlled stress exposure. Based on a modification of the Integrated Model of Anxiety and Perceptual-Motor Performance (Nieuwenhuys and Oudejans, 2017), we hypothesized an increase of psychophysiological stress reactions and a decrease of attention, decision-making and action quality.Methods: N = 59 German police officers (Mage = 34.16 years; SD = 7.79; Mwork experience = 11.33 years; SD = 9.04) completed a 3-hour training session in iVR with three increasingly stressful scenarios. Semi-structured, brief qualitative interviews were conducted directly after each scenario to assess attention, decision-making, and action. The interview questions were shown on a poster. Then participants, standing apart, recorded their oral answers via a tablet.Results: A total of 1,116 data units were extracted. Most responses referenced attention (n = 562), followed by action (n = 291) and decision-making (n = 263). Officers focused primarily on situational features, involved parties, and human factors that evoke stress. Decisions were often described as intuitive, with little reference to prior training content.Conclusion: The qualitative data offer unfiltered insights into officers’ immediate experiences and enrich the theoretical model. Combined with police guidelines, the findings help evaluate whether officers relied on task-relevant or irrelevant information during high-stress decision-making.
Marie Ottilie Frenkel, Jana Strahler
Open Access
Article
Conference Proceedings
Innovative MedEvac Decision, Coordination and Support System for Military Evacuation Scenarios
Modern battlefields are characterized by rapid redeployments, expansive operational zones, and evolving threats such as CBRN hazards, all of which render traditional medical support systems insufficient. The iMEDCAP project addresses these challenges by developing an integrated MedEvac Decision, Coordination, and Support System that revolutionizes battlefield casualty management. By leveraging autonomous technologies, the system incorporates advanced sensor networks – including wearable smart textiles and UAV-based multi-sensor reconnaissance – to continuously monitor soldiers’ physiological parameters and rapidly detect injuries. Upon casualty detection, the system activates a coordinated evacuation protocol managed by the Patient Evacuation Coordination Center (PECC), which dynamically allocates unmanned aerial and ground vehicles to transport patients using a specially designed patient transport module that ensures continuous remote monitoring and, if necessary, immediate medical intervention. Key innovations include real-time vital sign monitoring for early injury assessment, a decentralized approach to casualty detection and data integration, and autonomous transport solutions capable of short-, medium-, and long-distance evacuations. In addition, the project integrates diagnostic and intervention technologies that enable remote administration of first aid, significantly reducing the time between injury and critical treatment. This paper details the development, validation, and potential future enhancements of the iMEDCAP system, demonstrating its capacity to improve survival rates and operational efficiency in high-risk military scenarios. Through its user-centered design and robust decision support framework, iMEDCAP lays the groundwork for a next-generation European medical evacuation system, setting new standards in combat medical care.
Florian Haid, Anna Weber, Thomas Schnabel, Stefan Ladstätter, Julia Tschuden, Michael Schneeberger, Lucas Paletta, Alexander Almer, Markus Bergen, Gerald Bauer
Open Access
Article
Conference Proceedings
Real-Time Monitoring in Military Task Simulations: Insights from the RT-VitalMonitor Project
Modern military operations place soldiers under significant physical and cognitive stress, which can impact their performance and safety. The RT-VitalMonitor project addresses this challenge by developing a real-time monitoring system to assess the psychophysiological state of soldiers. This system utilizes advanced sensor technologies and data-driven models to continuously track critical physiological parameters such as heart rate, respiratory rate, and core body temperature, alongside performance metrics during demanding tasks. More than 170 participants from the Austrian Armed Forces underwent a series of infantry-specific tasks designed to replicate real-world combat conditions, including loaded marches, climbing over an obstacle, and casualty evacuation. The collected data provides valuable insights into soldiers' physical exertion, cognitive load, and overall readiness. The project's goal is to develop predictive models for task-specific load management, enabling a better understanding of soldiers' current states and forecasting their performance under varying levels of stress. The findings of this study contribute to more effective training strategies, enhanced operational safety, and the development of more personalized support systems for soldiers in high-stress environments. Ultimately, the RT-VitalMonitor project demonstrates the potential of wearable sensor technology and machine learning to revolutionize performance assessment and improve soldiers’ well-being in military operations.
Florian Haid, Julia Tschuden, Michael Schneeberger, Lucas Paletta, Anna Weber, Alexander Almer, Markus Bergen, Wolfgang Rausch, Gerald Bauer, Thomas Hölzl
Open Access
Article
Conference Proceedings
A Framework for Mixed Reality-supported Training of Conflict Resolution and First Responder Skills in International Crisis Situations: SmartSkills
Military first responders and also civilian experts in international peace missions must act quickly and unerringly in complex and dangerous situations under stress. Basic skills must be trained in advance as realistically as possible in order to be able to use them efficiently in later, real-life deployment situations and to reduce the risk potential when deployed to crisis areas. Simulations of concrete conflict scenarios are applied in the biannual “Native Challenge” workshop within a military camp area in the Austrian Alpes that is based on the idea of a cooperation between the “UNESCO Chair for Peace Studies” and the Military Command Tyrol. The representation of these simulations where participants experience in exercise scenarios the surprises, conflicts and dangers they can be confronted within a real mission is very costly which prohibits repeated, personalised and focused skill training.The Austrian project SmartSkills aims at providing standardised scenarios from the most diverse areas (behaviour at the checkpoint, negotiation, accident in the minefield, care of the wounded, etc.) interactively, tailored to the participants in terms of stress load, leadership ability and communication behaviour. A Mixed Reality training system allows unlimited repetitions, especially with personalised difficulty adaptations of the scenarios, and provides corresponding content for in-depth debriefings. SmartSkills will offer a highly innovative automated digital analysis of the human factors of the assignment, in particular the decisive situational awareness in critical situations, and uses biosensors to point out cognitive-emotional problem areas that require special attention in skill development. The associated Decision Support System will translate scientifically validated data into pragmatic risk estimates for the attention of the training management.This presentation will for the firstly describe the conceptual outline of the “Native Challenge” which integrates several levels and feedback loops of operational and strategical challenges. Secondly, the setting and the results of the initial requirement study detailed including information about the selected scenarios and use cases, with roles, objects of interest and training objectives. Furthermore, we will describe the technical specification for the Mixed reality system, its general system architecture, with the outline of wearable bio-signal sensors for psychophysiological monitoring, the graphical processing including realistic object visualisation and digital twins that were scanned from the real operational sites. We will provide details about the study plan, its research hypotheses, in particular, in the context of the evaluation of the training and psychophysiological key performance indicators. Finally, we will describe the anchoring in the international context and the acceleration of developments through exchange with European initiatives.SmartSkills is researching a new dimension of particularly realistic visualisations through the use of innovative digital twins: with highly accurate measurement technology and AI-supported evaluation software, internationally relevant environments can be experienced directly in the simulation centre, thereby increasing the realistic immersion.
Lucas Paletta, Andreas Peer, Markus Öttl, Georg Aumayr, K. Wolfgang Kallus, Joachim Brandtner, Gudrun Walter, Patrick Luley, Alexander Almer, Daniela Weismeier-Sammer, Christian Schönauer, Benjamin Schuster, Martin Söllner, Martin Müller, Michael Schneeberger, Martin Pszeida, Wolfgang Weiss, Anna Weber, Stefan Ladstätter, Florian Haid, Jochen A. Mosbacher, Silvia Russegger, Markus Bergen
Open Access
Article
Conference Proceedings
The Impact of Physical Strain on Performance in Basic Shooting Drills
There are many situations in the daily routine of military personnel and first responders where tasks that require a high degree of concentration and calm are performed during or immediately after strenuous physical activities. The resulting physical exhaustion and fatigue can affect performance in such tasks. While special training and general fitness might mitigate the impact on performance and outcome quality there still are open questions regarding the factors influencing the magnitude of an impact physical strenuous activity has and how such performance changes can be predicted.To get new insights, we analysed the shooting performance of 137 highly trained military personnel before and after a set of physically highly strenuous tasks. Questionnaires were used to assess, among other variables, amount of weekly physical training, pre-experience with the specific shooting drills, sleep quality, daily mood, and the rate of perceived exertion.The results indicated significant associations between the perceived exertion after the strenuous tasks and the change in the amount of false hits. Furthermore, there are correlations between the amount of physical training that participants conduct regularly and the amount of false hits after the strenuous tasks. Seeing false hits as an indicator of inhibitory processes, it is, from this kind of explorative study, conceivable that that good training status might mitigate detrimental effects of highly strenuous tasks on inhibitory ability. The study motivates for focusing future work on clarifying open questions and to provide more details.
Jochen A Mosbacher, Michael Schneeberger, Lucas Paletta, Martin Pszeida, Julia Tschuden, Anna Weber, Florian Haid, Silvia Russegger, Alexander Almer, Markus Bergen, Wolfgang Rausch, Gerald Bauer, Thomas Hölzl
Open Access
Article
Conference Proceedings
AI-driven Team Matching Using a Novel Personality Profile, Affinity Score and Fairness Measures
Research into the matching of employees in teams is crucial, as the success of a team depends heavily on the right composition. Individual skills, personalities and working styles can have a significant impact on team dynamics, collaboration and productivity. Targeted matching strategies can reduce conflicts, strengthen synergies and increase employee satisfaction, which ultimately promotes a company's innovative strength and competitiveness. Our research aims at the development of an e-tool that systemically matches people into existing teams based on their personality structure, professional role, and an algorithm for affinity mapping processes. We developed a novel methodology and decision support for recruitment being based on AI-based matching of personality profiles. We firstly apply a novel psychometric profile of personality structure being inspired by Häusl (2019), focusing on limbic-aligned dimensions: security and socialization, dominance and autonomy, stimulant and curiosity, challenge and risk, empathy and team, discipline and control. Furthermore, a quantitative collaboration assessment integrates features of communication, operation, relationship, and emotion, from the analysis of the interaction of personalities, defining unconscious and controlling patterns of thought and behavior and their impact on collaboration. The core components of our self-learning evaluation model include an algorithm that records and processes the relationships between personality and role system in the form of individual data and, derived from this, creates recommendations in the form of affinity matrices for the successful composition of teams. We define AI-driven assessment functions to determine the entries of an affinity matrix to be based on personality matching between a recruited and a given team. In a first step we apply a machine learning with explainable decision making that scores a cooperation potential based on the pairing of two personality profiles between 1 and 10 (maximum). We then seek for an optimal pairing based on a given team, starting with dyadic relation with team leaders, given a space of possible recruited that is developed with real and synthetic data. Finally, the affinity matrix is determined based on the recruited and compared to the optimally sampled individual. We present first results of the novel profile- and AI-based methodology. We recruited more than 90 personality profiles as well as 60 pairings, determined the cooperation score and applied expectation-maximization method (Dempster et al., 1977) to generate synthetic data to find optimally recruited. Fairness-based measures were applied in order to monitor potential bias in the distribution in the context of gender, such as, sex, age, or ethnic origin. Applying an ensemble of bagged trees (Zoghni, 2020) to estimate collaboration performance achieved a mean absolute error of 0.93 score points, using cross-correlation for training and test set data separation. Entrepreneurs and managing directors are facing complex challenges in human resources (HR) management; the pandemic has changed working morale and working models. Recruitment is time-consuming and becoming increasingly complex. The future target group are self-employed people and small and medium-sized enterprises. The application would range from individual tests or packages for the self-employed and small companies to a white-label service for large companies.
Michael Schneeberger, Jochen A Mosbacher, Fredrik Gustaffson, Georg Lindner Turecka, Wolfgang Weiss, Martin Pszeida, Thomas Orgel, Silvia Russegger, Lucas Paletta, Gerhard Handler, Cindy Luisser Haller
Open Access
Article
Conference Proceedings
CogDriver: The Longest-Running Autonomous Driving Cognitive Model Exhibits Human Factors
We are exploring how models can use models of human perception and motor control to interact directly with interfaces. We present CogDriver, a cognitive driving model capable of performing a long-duration autonomous driving task in a virtual simulation environment. This model, built using the ACT-R cognitive architecture and enhanced with robotic hands and eyes, supports the cognitive-perceptual-motor knowledge essential for simple human driving. It has two main strengths compared to other autonomous driving models: (a) it is built upon human-observed driving behavior, incorporating error-making and learning, and (b) it leverages a cognitive architecture to provide insights into psychological driving behavior. Compared to our previous version, this model shows improved endurance, maintaining its driving state for over 18 h from Tucson to Las Vegas, even under nighttime conditions. The enhancements were realized through incorporating human-like driving knowledge representations, and actions. It now includes a model of error handling and several logical visual cue strategies. The model's predictions can match certain aspects of human behavior in fine detail, such as the number of course corrections, average speed, learning rate, and adaptation to low visibility conditions. This model demonstrates that (a) perception and action loops with fallback handling provide a very accessible testbed for examining further aspects of behavior and (b) the model-task combination supports exploring aspects of human behavior that remain missing from ACT-R. Model, simulation, and data can be accessed at https://github.com/christianwasta/DriveBus/tree/drivebus-wasta.
Christian Wasta, Siyu Wu, Frank E. Ritter
Open Access
Article
Conference Proceedings
AI Audit as a tool for effective AI risk management
Artificial intelligence (AI), especially large language models (LLMs), is increasingly permeating all industries and promises transformative potential. Such large language models demonstrate impressive abilities in text generation, data analysis and even creative tasks. However, this rapid proliferation and increase in performance goes hand in hand with a growing awareness and concern about the manifold risks these technologies pose. The range of potential harms extends from operational malfunctions and data privacy breaches to profound systemic impacts on society and the economy. Given this duality of benefit and risk, there is an urgent need for robust governance, standardized risk management practices, and effective mitigation strategies. AI certification, specific risks associated with LLMs, corresponding mitigation techniques (technical and organizational), and the emerging concept of systemic AI risk. The standardization landscape for AI is still fragmented, but it is showing clear signs of convergence. AI audit using standards such as ISO 42001 and specific LLM risks support this process of security impact assessment.
Herwig Zeiner
Open Access
Article
Conference Proceedings
Fairness in Designing Decision-Making Processes with Multi-Agent Systems and Human Factors
This paper explores the integration of Human Factors (HF) into Multi-Agent Systems (MAS) to enhance fairness in decision-making processes in Industry 5.0 environments. We contribute with a human-centred perspective in the development of MAS by integrating the physiological aspects of workers in the manufacturing industry. This culminates in the measurement of human resilience. This paper presents an automotive manufacturing environment where wearable sensors and AI-driven analytics assess workers' physiological and psychological stress levels to calculate a human resilience score. This score, along with worker preferences, supports a dynamic worker allocation algorithm based on MAS that adapts to production demands. Our approach embodies the Industry 5.0 vision of technologies that support adaptive, transparent and, above all, fair human management. The system uses advanced technologies to meet business goals and employee needs and to promote a more inclusive, supportive and people-centred work environment.
Gustavo Vieira, Herwig Zeiner, Lucas Paletta, Rui Fernandes, Higor Rosse, Khadija Sabiri, Juan F De Paz, José Barbosa
Open Access
Article
Conference Proceedings
The Algorithmic Fairness Challenge in Decision-Making
The increasing use of automated decision-making systems has given rise to concerns about fairness. This paper examines the major principles of fairness in decision-making, and it discusses the challenges of implementing fairness principles in practice, such as the trade-offs between different types of fairness and the difficulty of measuring fairness. Finally, the paper puts forward several proposals for promoting fairness in decision-making. These include the use of transparent and explainable algorithms, the involvement of stakeholders in the design of decision-making systems, and the establishment of accountability mechanisms. It is of the utmost importance that fairness is a fundamental principle in decision-making processes. This approach is designed to ensure that individuals facing similar circumstances are treated equally and not subjected to discrimination. Examples of unfair decision-making include situations where individuals are discriminated against based on protected attributes such as race, gender, or age. Decisions that lack transparency in their process may be perceived as biased or unjust. Unfairness can arise in a number of ways, for example when promotions are based on favouritism rather than merit, or when hiring decisions are influenced by personal biases rather than qualifications.
Herwig Zeiner, Lucas Paletta, Gustavo Vieira
Open Access
Article
Conference Proceedings