Hibridni događaj

Program događaja

četvrtak, 28.5.2026 8:00 - 13:00, Bellavista, Grand hotel Adriatic, Opatija Radovi   1. D. Kim (Texas Christian University, Fort Worth, United States), J. Valacich (University of Arizona, Tucson, United States), J. Jenkins, D. Wilson (Brigham Young University, Provo, United States) Trace Data as a Privacy-Preserving Indicator of Compliance Behavior in External Generative AI Use  The risks of organizational compliance violations may increase as employees route digital work tasks to external generative AI (GenAI) tools. When users transfer work artifacts into prompts with little intervening review or revision—often via copy–paste—the likelihood of inadvertently exposing sensitive information (e.g., confidential firm data or personally identifiable information) may rise. While monitoring and auditing solutions can track outbound information flows, they are often costly and may raise privacy concerns. This exploratory study examines trace data–derived digital behavioral biometric measures that characterize prompt-entry mechanics during GenAI composition and submission. Using an instrumented web interface with client-side event logging, we compare interaction traces across two simulated conditions: a baseline condition that permits direct copy–paste transfer and a secure-instruction condition that instructs users to redact sensitive information prior to submission. Across measures anchored to the paste-to-submit window, the secure instruction condition shows reduced paste reliance and, among paste users, longer paste-to-submit delays, more postpaste editing activity, and longer paste-to-first-edit latencies. These patterns suggest that interaction-log–based measures can non-intrusively distinguish rapid transfer-and-submit behavior from transformation-oriented editing without inspecting prompt content. 2. A. Villarreal, A. Sachdeva, L. Brandimarte (University of Arizona, Tucson, United States) LLM Randomness and Creativity: An Experiment on Human-AI Collaboration  Large Language Models (LLMs) can boost human creativity in a variety of tasks. Little is known, however, about how various customizable parameters of a given LLM affect the creativity of the human-AI collaboration outcome. In this work, we focus on the temperature parameter (which affects the randomness of the LLM suggestion) and investigate its impact on two main different types of human creativity: convergent and divergent thinking. We hypothesize that low randomness boosts convergent thinking creativity, while high randomness enhances divergent thinking. We test this hypothesis through a 3x3 full factorial between-subject experiment, in which we manipulate LLM randomness and type of task, and measure outcome creativity. The temperature parameter is set to 0.1 in the low randomness condition and 1.2 in the high randomness condition. To these two levels, we add a control condition in which participants do not get the assistance of a LLM. For the task manipulation, participants get randomly assigned to either complete the Remote Associates Test (which measures convergent thinking) or one between the Alternate Uses Test and the Divergent Association Task (which measures divergent thinking). We also investigate the role of human engagement with the LLM via detailed telemetry captured from the human-LLM interaction. 3. M. Juric (University of Zadar, Zadar, Croatia) Exploring the Roles of Machine Agency and Machine Heuristic in User Experience and Engagement with Generative AI  As university students increasingly adopt generative AI, traditional user experience models struggle to account for the complexity of their engagement. This study examines how machine agency, the perception of the system as an autonomous, intelligent collaborator, shapes pragmatic and hedonic dimensions of user experience, mobilizes focused attention, and ultimately contributes to a sense of reward. The machine heuristic, a cognitive shortcut that attributes objectivity to automated sources, is examined as a parallel explanatory mechanism. Three structural equation models were compared on a sample of 146 university students: a pragmatic-heuristic model, a social-affective model, and an integrated model. The integrated framework proved most effective, accounting for nearly 80% of variance in perceived reward. Agency operates through fully mediated independent pathways rather than through the hypothesized serial chains. Specifically, the expected serial mediation from agency through hedonic quality to focused attention did not hold; instead, agency directly mobilized attention. Similarly, the machine heuristic did not predict pragmatic quality as anticipated but acted as a direct shortcut to reward. Focused attention, pragmatic quality, and the machine heuristic thus function as three independent mediators between perceived agency and reward. These results position machine agency as the primary driver of student engagement with generative AI. 4. K. Hape, C. Milewicz (University of Southern Indiana, Evansville, United States) Advancing Student Skills in Therapeutic Use of Self and Gerontology Using Artificial Intelligence and Biometric Feedback  Artificial Intelligence (AI) is evolving and changing the face of higher education in transformative ways by offering intuitive opportunities for personalized student feedback. Key opportunities are available to harness emerging technologies and increase student growth in client interaction skills utilizing AI-human interactions. This research demonstrates how AI powered facial expression analysis (FEA) can be used to teach occupa-tional therapy students to apply therapeutic use of self (TUS) in computer-mediated telehealth treatment. Emo-tions are measured using emotion AI technology for FEA. Students' telehealth videos were recorded, and FEA was applied to identify both client and student non-verbal communication. This process allowed students to see how their words produced joy in their client and how joy was mirrored between the student and the client. Student survey results support the integration of biometric feedback when learning TUS and indicate perceived growth in student TUS skills. 5. T. Orehovački (Juraj Dobrila University of Pula, Pula, Croatia) What Drives Students’ Continued Use of Large Language Models for Programming Tasks?  Large language models (LLMs) are increasingly embedded in students’ programming workflows, supporting activities such as code generation, debugging, and concept explanation. However, the educational value of LLM assistance depends not only on the availability of generated solutions, but also on whether students perceive the outputs as dependable and whether interaction with the tool supports cognitively demanding engagement with code. This study proposes and validates a parsimonious model explaining students’ intention to continue using LLMs for programming tasks through the interplay of response reliability, critical thinking development, perceived code quality, and learning effectiveness. Data were collected via a structured questionnaire administered to university students and analyzed using partial least squares structural equation modeling (PLS-SEM). The results show that response reliability and critical thinking development significantly contribute to perceived code quality, while response reliability and code quality positively influence learning effectiveness. Learning effectiveness, in turn, emerges as the strongest predictor of behavioral intention to continue using LLMs for programming-related learning and task execution. These findings suggest that sustained adoption of LLMs in programming education is shaped by a chain of perceptions that connect dependability and critical engagement to code-related outcomes and perceived learning progress. The study provides a concise explanation of continued LLM use grounded in programming-specific benefits and highlights the importance of reliability and verification-oriented engagement in LLM-assisted programming learning. 6. J. Gjurčević, P. Kolar (VERN' University, Zagreb, Croatia) A Comparative Analysis of Conversational Artificial Intelligence Models for Python Software Development  The effectiveness of conversational artificial intelligence models in the development of a Python application for calculating simple interest is examined in this paper. The study investigates the capabilities of seven conversational models (ChatGPT, Gemini, Copilot, Claude, MetaAI, DeepSeek, and Perplexity) in developing a Python application with a Tkinter interface for simple interest calculation. A double evaluation was conducted in both English and Croatian. Data were collected on the number of prompts required to obtain a functional solution, code length, and user interface quality. Quantitative analysis was applied (number of prompts, lines of code, criterion-based scoring), along with a qualitative assessment of usability. The results indicate higher stability and efficiency when prompts were issued in English. The best results were achieved by Claude and Gemini, which produced functional solutions with the fewest prompts and high overall scores. In the Croatian-language evaluation, Perplexity, alongside Claude, achieved the best results, while ChatGPT demonstrated consistent performance in both languages. A negative correlation was observed between the number of prompts and solution quality, whereas the number of lines of code showed only a moderate correlation with the overall score. Issues of consistency were identified, such as MetaAI’s failure to produce a functional solution in Croatian. The study concludes that language choice is a critical factor and that conversational models differ in their balance between simplicity and functionality. 7. Z. Vidačković (Croatian National Theatre Zagreb, Zagreb, Croatia) Human–AI Interaction in Film Production: A Communication Perspective  Research on artificial intelligence in film production is often fragmented across individual domains such as scriptwriting, visual effects, editing, or marketing, without adequately conceptualising film production as a continuous human-AI interaction process. This paper addresses that gap by proposing a lifecycle framework of human-AI interaction across the major phases of film production: development, pre-production, production, post-production, distribution, and audience engagement. Drawing on a human-computer interaction perspective and on the understanding of film as a mass communication medium, the paper models each phase as a structured interaction environment involving human creators, AI system functions, and specific forms of technological mediation. Within this framework, AI is interpreted not as an autonomous creative agent but as a distributed decision-support, optimisation, and generative-assistance layer embedded in human-centred workflows. The paper identifies recurring patterns of shared decision-making, user agency, human oversight, and communicative mediation across the production lifecycle, and relates them to broader issues of authorship, responsibility, transparency, and ethical design. Rather than offering empirical evaluation of a single AI system, the study makes a conceptual contribution by formalising a structured model for analysing how human and AI roles interact throughout contemporary film production. In this way, the paper contributes to HCI and media production research by providing a clearer analytical framework for understanding AI as a collaborative but subordinate actor within film as a mass-media environment. 8. M. Gombar (Ministry of Defense of the Republic of Croatia, Zagreb, Croatia), M. Pejić Bach (Faculty of Economics and Business, University of Zagreb, Zagreb, Croatia; Faculty of Commercial and , Zagreb, Croatia), M. Omazić (Faculty of Economics and Business, University of Zagreb, Croatia; IEDC – Bled School of Management, , Zagreb, Croatia) Mapping Human–Algorithm Interaction: A Bibliometric and Conceptual Map of Transparency, Autonomy, and Digital Resilience (2010–2026)  Algorithms increasingly shape how people allocate attention, make decisions, and participate in digital environments. As organizations rely ever more heavily on algorithmic systems to create, scale, and capture business value, understanding how these systems function from the perspective of their use becomes increasingly important. Despite their growing influence, the conceptual vocabulary used to describe human–algorithm interaction remains scattered across disciplines. This study consolidates key concepts and systematically maps the research landscape of human–algorithm interaction within the HCI field from 2010 to 2026, based on a Scopus corpus of 377 publications. Using a custom-built thesaurus, we apply VOSviewer to generate three complementary bibliometric networks: keyword co-occurrence based on author keywords, source co-citation at the journal level, and bibliographic coupling at the document level. An international co-authorship map is added, while Bibliometrix is used to examine conceptual structures and thematic evolution over time. These include large language models and generative AI, with growing attention to retrieval-augmented generation and knowledge graphs; transparency and explainability, closely tied to fairness and interpretability; recommender systems, with a strong emphasis on trust; and human–algorithm interaction and user experience, particularly in relation to user control and responsibility. Overlay views indicate a clear post-2021 shift toward LLM-centric topics. The co-citation spine links HCI venues (CHI/TOCHI; Computers in Human Behavior) with ML/AI outlets (NeurIPS, JMLR, Information Fusion), evidencing convergence. We distill a practical pathway: transparency → autonomy → resilience, and propose resilience-by-design patterns for LLM and recommender interfaces (clear provenance, controllable personalization, exposure diversity, salient error signaling). A reproducibility package (thesaurus and VOS configurations) supports robustness and sensitivity checks. Overall, this work offers a consolidated conceptual map and a design-oriented research agenda aimed at strengthening cognitive autonomy, digital resilience, and the responsible creation of business value within HCI. 9. R. Hensley, J. Wilcox, P. Suszko, S. Mesones, D. Bačić (Loyola University Chicago, Chicago, United States) Where Notifications Appear Matters: Eye-Tracking Evidence of Spatial Effects on Attention in Workplace Interruptions  Digital notifications are a pervasive feature of modern knowledge work, yet their disruptive impact depends not only on their presence but on how they are presented within the interface. This study investigates how notification placement and auditory signaling influence visual attention during an office-style task. Thirty-four participants completed timed email-response trials while Microsoft Teams notifications appeared under five conditions: four visual placements (top-left, top-right, bottom-left, bottom-right) and a bottom-right notification with sound. Eye tracking captured fixation count, fixation duration, dwell time, and revisit behavior within notification areas of interest. Results reveal two distinct attentional mechanisms. Notifications appearing on the left side of the interface elicited more frequent orienting responses, reflected in higher fixation counts and revisits. In contrast, notifications appearing in the bottom regions produced longer fixation durations and greater dwell time, indicating deeper cognitive processing once attention was captured. The addition of an auditory alert did not significantly affect any attention metric relative to the default bottom-right visual notification. These findings demonstrate that spatial placement is a primary determinant of how notifications capture and engage attention, exceeding the influence of modality. The results position notification placement as a critical and underexplored design parameter in managing interruptions in workplace communication systems. 10. M. Gregurić, E. Ivanjko (University of Zagreb, Faculty of Transport and Traffic Sciences, Zagreb, Croatia), N. Hlupić (University of Zagreb, Faculty of Electrical Engineering and Computing, Zagreb, Croatia), D. Čakija (University of Zagreb, Faculty of Transport and Traffic Sciences, Zagreb, Croatia) Accountability of Cognitive Processes Extracted from Video Game Players in the Development of Transport and Combat Autonomous Vehicles  Video games in which human players control various vehicles to achieve specific goals can serve as a potential simulation environment for tracking and learning human control patterns. This study proposes a conceptual framework that can utilize video games as a training environment for the Artificial Intelligence (AI) based bots that control in-game vehicles in interactions with human players. The concept uses trained AI models of bots that interact with human players during the initial training steps for autonomous vehicles for transport and combat tasks. The transferred models from in-game AI bots can greatly stabilize learning convergence and speed up the overall training of real-world autonomous vehicles. Thus, this study addresses the problem of accountability and possible control mechanisms in generating desirable behavioural patterns of human players in controlling in-game transport and combat vehicles. Special focus is set on cognitive distortions within players’ behavioral patterns, which can hinder their accountability for transfer learning. 11. L. Žužić, J. Lerga (University of Rijeka, Faculty of Engineering, Rijeka, Croatia) Enhancing greenhouse gas emission awareness through web-based tools  This paper presents a web-based human-computer interaction (HCI) experiment designed to measure the educational efficacy of a card game using a 32-card "Top Trumps" (or "War") mechanic in which each card represents a country and its 2024 greenhouse gas (GHG) emissions, sourced from the Emissions Database for Global Atmospheric Research (EDGAR) report. A key component of the research design is the deck, a curated set of countries selected for their high GHG emissions per unit of gross domestic product (GDP), which focuses the player's attention on global carbon inefficiency and emissions intensity. The study employs an 8-stage experimental design where participants first view the data, take a quiz, and then play the game against a deterministic heuristic algorithm. This cycle is then repeated to measure changes in knowledge retention. We evaluate the game's impact by comparing pre- and post-game quiz scores and by analyzing user-reported cognitive load and engagement using a National Aeronautics and Space Administration Task Load Index (NASA-TLX) survey. This research provides a framework for evaluating gamified learning models for complex, real-world data and assesses the strategic interaction between a human player and a simple, data-driven algorithm. 12. M. Vukšić, J. Ćelić, I. Panić, L. Liker (Sveučilište u Rijeci, Pomorski fakultet, Rijeka, Croatia) Human-in-the-Loop Digital Twin Monitoring for Refrigerated Container Maintenance  Failures of refrigerated containers during maritime transport represent a significant operational risk, as even short-term loss of temperature control can result in substantial cargo damage. This paper presents a human-in-the-loop digital twin monitoring approach for supporting predictive maintenance of refrigerated containers through early fault detection and operator-centred decision support. A physics-based digital twin continuously estimates the expected thermal and operational behaviour of a container using real-time sensor data. Deviations between predicted and observed behaviour are analysed using a data-driven anomaly detection model trained on normal operating conditions. To ensure usability in safety-critical maritime environments, analytical results are integrated into a mixed reality (MR) interface explicitly designed according to human–computer interaction principles, emphasising transparency, selective alerting, interpretability, and operator trust rather than autonomous control. The approach is evaluated using laboratory experiments on a REFCON-based testbed, supported by high-fidelity simulations and synthetically generated degradation data. The results show that cooling performance degradation and sensor-related anomalies can be detected at an early stage while maintaining stable behaviour during nominal operation. The findings indicate that combining digital twin analytics with intuitive human–computer interfaces can enhance predictive maintenance by supporting human situation awareness and informed decision-making in maritime refrigerated transport. 13. M. Duffy, F. Saiyed, A. Tokarczyk, D. Bačić (Loyola University Chicago, Chicago, United States) Trust, Attention, and Moderation Architecture: Comparing Platform-Driven and Community-Driven Warning Labels  Social media platforms increasingly deploy warning labels to mitigate the impact of misinformation; however, limited empirical research examines how the source framing of moderation influences user perception and attention. This study investigates whether platform-driven and community-driven warning messages differentially affect trust in the warning, confidence in content accuracy judgments, and visual attention allocation. Using a within-subject laboratory experiment (N = 39), participants viewed political and health-related misinformation posts under both warning conditions while eye-tracking and survey data were collected. Results indicate that community-driven warnings significantly increase perceived trust and redistribute visual attention toward the warning area of interest. However, confidence in content accuracy judgments did not differ significantly across conditions, suggesting that increased trust does not necessarily translate into belief revision. The findings highlight the importance of moderation architecture in shaping both evaluative perception and perceptual processing, offering implications for the design of transparent and participatory misinformation interventions. 14. P. Tomšič, I. Vasileska (Faculty of Mechanical Engineering, University of Ljubljana, Ljubljana, Slovenia) Educational Activities and Lessons Learned from the EuroCC 2 National Competence Center for Supercomputing in Slovenia  Over the past three years, the EuroCC 2 project has played a key role in establishing and strengthening the National Competence Center (NCC) for supercomputing in Slovenia, with a strong emphasis on education and training in high-performance computing (HPC). This paper presents an overview of the educational activities conducted within the Slovenian NCC from 2023 to 2025 and analyzes the lessons learned from their implementation. The analysis is based on quantitative data collected from training events, including the total number of courses, participant attendance, and the distribution of domestic and international participants. Particular attention is given to trends in participation over time and to the effectiveness of different training formats. Furthermore, the paper examines the impact of collaborations with academic institutions, research organizations, and international EuroHPC partners on the reach and quality of the training program. The results indicate that targeted collaborations significantly increased participation, especially among early-career researchers and students, and contributed to a higher share of foreign attendees. The findings highlight the importance of sustained partnerships, adaptive training content, and community-oriented approaches in building national HPC competencies. The presented lessons learned may serve as a reference for other national competence centers and educational initiatives in the field of supercomputing. 15. R. Rubčić, L. Grediček (VERN' University, Zagreb, Croatia) Development and Security Testing of a Children Smart Toy Model  This paper presents the development and security testing of a prototype children smart toy from a human computer interaction perspective with emphasis on trust safety and parental perception. Children smart toys represent highly interactive systems that combine physical interaction voice communication and network connectivity. While these features enhance engagement they also introduce risks that directly affect user trust and perceived safety. A functional prototype was developed on the Raspberry Pi platform and includes speech based interaction audio playback image capturing and storytelling. To reflect realistic usage scenarios several security weaknesses were intentionally introduced during development. Security testing was conducted in a controlled local network environment using passive traffic monitoring and active attack techniques. The results revealed vulnerabilities related to authentication data protection and unsecured communication channels that may negatively influence parental trust and acceptance of such systems. A comparison with publicly reported incidents involving commercial smart toys showed strong similarities in both technical shortcomings and potential user impact. The findings demonstrate that security is not only a technical requirement but a key factor influencing interaction quality trust and responsible design of children oriented interactive systems. 16. D. Šturlan, R. Mohorović, T. Orehovački (Sveučilište Jurja Dobrile u Puli, Pula, Croatia) Korisničko iskustvo i obrasci korištenja film streaming aplikacija među hrvatskim studentima  Razvoj film streaming aplikacija promijenio je način konzumacije audiovizualnog sadržaja, pri čemu korisničko iskustvo ima važnu ulogu u prihvaćanju i dugoročnom korištenju tih platformi. U ovom su radu analizirani obrasci korištenja film streaming aplikacija među hrvatskim studentima te zadovoljstvo povezano s estetikom korisničkog sučelja, funkcionalnošću, ponudom sadržaja i ukupnim korisničkim iskustvom. Istraživanje je provedeno putem online ankete na uzorku studenata prijediplomskih i diplomskih studija. Rezultati analize prikupljenih podataka potvrdili su dominaciju Netflixa kao preferirane film streaming aplikacije među ispitanicima. Razlike s obzirom na spol utvrđene su u kontekstu učestalosti njihova korištenja, dok zadovoljstvo ponudom sadržaja nije bilo povezano s tom varijablom. Usporedba Netflixa i HBO-a nije pokazala statistički značajne razlike u doživljaju estetike korisničkog sučelja. Istodobno, ponuda sadržaja imala je veću težinu u oblikovanju iskustva korištenja od vizualnih obilježja sučelja. Prezentirani rezultati mogu poslužiti kao uporište za daljnja istraživanja korisničkog iskustva film streaming aplikacija te kao smjernice za unaprjeđenje njihova dizajna i ponude sadržaja. 17. N. Nelufule, Pretoria, South Africa), N. Siphambili, Pretoria, South Africa), D. Shadung, Pretoria, South Africa), P. Senamela (Council for Scientific and Industrial Research (CSIR), Pretoria, South Africa) Workforce Skills Gaps and Human-AI Collaboration in Adaptive Factories  The transition from Industry 4.0 to Industry 5.0 has repositioned humans at the center of adaptive, resilient, and sustainable manufacturing systems. It has been projected that by then end of November 2025, over 68 % of global manufacturers will report lack of critical workforce skills that also impede full adoption of AI-enabled adaptive factories. In this article, a survey results on the nature, magnitude, and evolution of the lack of critical workforce skills, and the emerging paradigms of human-AI collaboration are presented. A PRISMA framework was used to synthesize peer-reviewed articles between 2020 to 2026 to examine the existing dominant themes, ranging from technical deficiencies in AI literacy and data science to socio-emotional and creative skills required for effective robot interaction. The main research contribution in this article is the Human-AI Synergy Competency Framework, which is a multilevel, dynamic model that maps required competencies, assesses maturity, and prescribes personalized reskilling pathways using generative AI tutors and digital twins. This research has also revealed that current AI tutoring technologies have demonstrated faster upskilling of about 57 % and 28–54 % of productivity gains based on the simulated data. This article has also recommended the adoption of regulatory mandates particularly for the lifelong learning credits and enterprise adoption of Human-AI Synergy Competency Framework frameworks to reduce the projected global manufacturing talent shortfall of 8.5 million workers, by 2030.

The student discount doesn't apply to PhD students.

NOTE FOR AUTHORS: In order to have your paper published, it is required that you pay at least one registration fee for each paper. Authors of 2 or more papers are entitled to a 10% discount.

Joe Valacich
University of Arizona
Eller College of Management
McClelland Hall Rm 430L, PO Box 210108
Tucson, AZ 85721, United States

E-mail: valacich@arizona.edu 
 

Dario Ogrizovic
University of Rijeka
Faculty of Maritime Studies
Studentska 2
HR-51000 Rijeka, Croatia

E-mail: dario@uniri.hr
 

The best papers will get a special award.
Accepted papers will be published in the ISSN registered conference proceedings. Papers presented at the conference will be submitted for inclusion in the IEEE Xplore Digital Library.

Opatija is the leading seaside resort of the Eastern Adriatic and one of the most famous tourist destinations on the Mediterranean. With its aristocratic architecture and style, Opatija has been attracting artists, kings, politicians, scientists, sportsmen, as well as business people, bankers and managers for more than 180 years.

The tourist offer in Opatija includes a vast number of hotels, excellent restaurants, entertainment venues, art festivals, superb modern and classical music concerts, beaches and swimming pools – this city satisfies all wishes and demands.

Opatija, the Queen of the Adriatic, is also one of the most prominent congress cities in the Mediterranean, particularly important for its ICT conventions, one of which is MIPRO, which has been held in Opatija since 1979, and attracts more than a thousand participants from over forty countries. These conventions promote Opatija as one of the most desirable technological, business, educational and scientific centers in South-eastern Europe and the European Union in general.

For more details, please visit www.opatija.hr and visitopatija.com.