Within journalistic editorial processes, disclosing AI usage is currently limited to simplistic labels, which misses the nuance of how humans and AI collaborated on a news article. Through co-design sessions (N=10), we elicited 69 disclosure designs and implemented four prototypes that visually disclose human–AI collaboration in journalism. We then ran a within-subjects lab study (N=32) to examine how disclosure visualizations (Textual, Role-based Timeline, Task-based Timeline, Chatbot) and collaboration ratios (Primarily Human vs. Primarily AI) influenced visualization perceptions, gaze patterns, and post-experience responses. We found that textual disclosures were least effective in communicating human-AI collaboration, whereas Chatbot offered the most in-depth information. Furthermore, while role-based timelines amplified AI contribution in primarily human articles, task-based timeline shifted perceptions toward human involvement in primarily AI articles. We contribute Human-AI collaboration disclosure visualizations and their evaluation, and cautionary considerations on how visualizations can alter perceptions of AI’s actual role during news article creation.

We present a multi-modal stress dataset that uses digital job interviews to induce stress. The dataset provides multi-modal data of 40 participants including audio, video (motion capturing, facial landmarks, eye tracking), as well as physiological information (photoplethysmography, electrodermal activity). In addition to that, the dataset contains time-continuous annotations for stress and occurred emotions (e.g., shame, anger, anxiety, and surprise). In order to establish a baseline, five different machine learning classifiers (Support Vector Machine, K-Nearest Neighbors, Random Forest, Feed-forward Neural Network, and Long-Short-Term Memory Network) have been trained and evaluated on the presented dataset for a binary stress classification task. The best-performing classifier has been a Long-Short-Term Memory Network, which achieved an accuracy of 91.7% and an F1-score of 90.2%. The ForDigitStress dataset is freely available to other researchers.

Automatic stress detection using heart rate variability (HRV) features has gained significant traction as it utilizes unobtrusive wearable sensors measuring signals like electrocardiogram (ECG) or blood volume pulse (BVP). However, detecting stress through such physiological signals presents a considerable challenge owing to the variations in recorded signals influenced by factors, such as perceived stress intensity and measurement devices. Consequently, stress detection models developed on one dataset may perform poorly on unseen data collected under different conditions. To address this challenge, this study explores the generalizability of machine learning models trained on HRV features for binary stress detection. Our goal extends beyond evaluating generalization performance; we aim to identify the characteristics of datasets that have the most significant influence on generalizability. We leverage four publicly available stress datasets (WESAD, SWELL-KW, ForDigitStress, VerBIO) that vary in at least one of the characteristics such as stress elicitation techniques, stress intensity, and sensor devices. Employing a cross-dataset evaluation approach, we explore which of these characteristics strongly influence model generalizability. Our findings reveal a crucial factor affecting model generalizability: primary stressor. Models achieved good performance across datasets when the primary stressor (e.g., social evaluation in our case) remains consistent. Factors like stress intensity or brand of the measurement device had minimal impact on cross-dataset performance. Based on our findings, we recommend matching the primary stressor when deploying HRV-based stress models in new environments. Although previous works have performed cross-dataset evaluation of stress models, this is the first study to systematically investigate the factors influencing the cross-dataset applicability of HRV-based stress models. Our insights are crucial for scenarios with limited data, where techniques like domain generalization and domain adaptation may not be applicable.

This study evaluates a socially interactive agent to create an embodied cobot. It tests a real-time continuous emotional modeling method and an aligned transparent behavioral model, BASSF (boredom, anxiety, self-efficacy, self-compassion, flow). The BASSF model anticipates and counteracts counterproductive emotional experiences of operators working under stress with cobots on tedious tasks. The flow experience is represented in the three-dimensional pleasure, arousal, and dominance (PAD) space. The embodied covatar (cobot and avatar) is introduced to support flow experiences through emotion regulation guidance. The study tests the model's main theoretical assumptions about flow, dominance, self-efficacy, and boredom. Twenty participants worked on a task for an hour, assembling pieces in collaboration with the covatar. After the task, participants completed questionnaires on flow, their affective experience, and self-efficacy, and they were interviewed to understand their emotions and regulation during the task. The results suggest that the dominance dimension plays a vital role in task-related settings as it predicts the participants' self-efficacy and flow. However, the relationship between flow, pleasure, and arousal requires further investigation. Qualitative interview analysis revealed that participants regulated negative emotions, like boredom, also without support, but some strategies could negatively impact well-being and productivity, which aligns with theory.

Stress is prevalent in many aspects of everyday life including work, healthcare, and social interactions. Many works have studied handcrafted features from various bio-signals that are indicators of stress. Recently, deep learning models have also been proposed to detect stress. Typically, stress models are trained and validated on the same dataset, often involving one stressful scenario. However, it is not practical to collect stress data for every scenario. So, it is crucial to study the generalizability of these models and determine to what extent they can be used in other scenarios. In this paper, we explore the generalization capabilities of Electrocardiogram (ECG)-based deep learning models and models based on handcrafted ECG features, i.e., Heart Rate Variability (HRV) features. To this end, we train three HRV models and two deep learning models that use ECG signals as input. We use ECG signals from two popular stress datasets WESAD and SWELL-KW - differing in terms of stressors and recording devices. First, we evaluate the models using leave-one-subject-out (LOSO) cross-validation using training and validation samples from the same dataset. Next, we perform a cross-dataset validation of the models, that is, LOSO models trained on the WESAD dataset are validated using SWELL-KW samples and vice versa. While deep learning models achieve the best results on the same dataset, models based on HRV features considerably outperform them on data from a different dataset. This trend is observed for all the models on both datasets. Therefore, HRV models are a better choice for stress recognition in applications that are different from the dataset scenario. To the best of our knowledge, this is the first work to compare the cross-dataset generalizability between ECG-based deep learning models and HRV models.

Emotion expression is an important part of human-robot interaction. Previous studies typically focused on a small set of emotions and a single channel to express them. We developed an emotion expression model that modulates motion, poses and LED features parametrically, using valence and arousal values. This model does not interrupt the task or gesture being performed and hence can be used in combination with functional behavioural expressions. Even though our model is relatively simple, it is just as capable of expressing emotions as other more complicated models that have been proposed in the literature. We systematically explored the expressivity of our model and found that a parametric model using 5 key motion and pose features can be used to effectively express emotions in the two quadrants where valence and arousal have the same sign. As paradigmatic examples, we tested for happy, excited, sad and tired. By adding a second channel (eye LEDs), the model is also able to express high arousal (anger) and low arousal (relaxed) emotions in the two other quadrants. Our work supports other findings that it remains hard to express moderate arousal emotions in these quadrants for both negative (fear) and positive (content) valence.