Abstract Assistive technologies are increasingly being developed to support the diverse cognitive, social, and sensory needs of neurodivergent individuals to improve their quality of life. Technologies such as virtual reality (VR), mobile applications, and smartwatches facilitate participation in daily activities, education, work, and social interactions. This paper presents a systematic literature review of 38 articles examining assistive technologies designed for neurodivergent individuals, with the aim of identifying technological advances and their impact on independence and quality of life. Following the PRISMA framework, papers published between 2018 and 2024 were retrieved from the IEEE Xplore and ACM Digital Library databases using neurodiversity-oriented terminology. The findings highlight how assistive technologies such as mobile applications, VR headsets, and smartwatches support daily activities including social interaction, physical exercise, routine management, and self-regulation. The review also underscores the importance of participatory and co-design approaches in technology development and identifies a significant research gap regarding the needs of older neurodivergent populations, calling for further investigation in this area.
Abstract Neurodivergent employees, encompassing those with conditions like attention deficit/hyperactivity disorder, autism spectrum disorder, and giftedness, offer unique cognitive strengths yet face challenges in neurotypical environments. This article presents CoDiT-AI, a framework integrating digital twins (DTs) and artificial intelligence (AI) to create workplaces optimized for neurodivergent individuals. It introduces the concept of a Neuro-Workplace-Twin, a digital representation that simulates neurodivergent sensory experiences and employs an AI-based perceptual model to transform workplace designs, thereby supporting the design of inclusive workspaces. Additionally, AI technologies, particularly large language models like GPT-4, are utilized to bridge communication gaps and to identify tasks. The combination of DTs and AI technologies supports the transformation of workplaces into neurodivergent-friendly environments. Empirical data from workshops, interviews, and surveys highlight key needs and preferences for adaptations in the workplace and can be used as a basis for the DT.
Despite their increasing adoption, understanding black-box models that underpin AI technologies is not trivial. To address this challenge, explainable AI (XAI) tools have received increased interest, albeit their HCI exploration is limited. We report a systematic review of 142 papers focused on the design, use, or evaluation of XAI tools, and in particular, the types of XAI tools, their exploration, the type of data input and output, the type of explanation, available XAI interfaces, and evaluation of XAI tools. Our outcomes highlight a large number of XAI tools, albeit a small number is commonly used, most often by AI experts as users rather than novice users. Our findings led to four research, design, and methodological implications, including the call for more HCI-driven user interaction with XAI tools, for designing richer portfolios of user-data interaction scenarios for XAI tools, and of novel XAI user interfaces, and for leveraging hybrid methods and measures for XAI tools evaluation.
Gender stereotypes can influence user interactions with digital agents, shaping perceptions and behaviors based on avatar representation. However, gender-neutral designs have only recently emerged as a potential strategy to challenge and reduce these biases. Moreover, the effectiveness of such designs remains largely underexplored in the context of non-binary individuals, who are often excluded from evaluation studies due to their low representation. In an online survey experiment using a 3x3 mixed factorial design and a validated set of 9 stimuli, participants of different self-reported genders evaluated masculine, feminine, and gender-neutral avatar faces across multiple perceptual dimensions (likeability, agency, communion, interaction, and attractiveness), together with measures of ambivalent sexism (ASI) and ambivalence toward men (AMI). The results revealed a strong effect of participant gender on avatars perception, with non-binary participants providing significantly lower ratings than both men and women for all dependent variables. When gender-prejudice measures were included, these group differences were substantially reduced, revealing that dimensions of ambivalent sexism and ambivalence toward men shaped avatar perception. Benevolence toward men increased ratings only among male participants, whereas hostility toward men enhanced evaluations across all participants. Hostile sexism also interacted with both participant and avatar gender, influencing agency and communion judgments. Overall, the results indicate that rather than pursuing the search for seemingly gender-neutral technologies, it is more appropriate to include in the design processes of these technologies, participants with different genders and different attitudes toward the multiplicities of gender expression, and to employ more gender-diverse rather than gender-neutral avatar representations.
Artificial intelligence (AI)-supported automated triage systems that incorporate auto-anamnesis, algorithmic triage, and text-based communication with healthcare professionals are increasingly deployed to optimize workflows and expand access to care. However, their influence on the work engagement and psychological wellbeing of healthcare professionals remains underexplored. Using self-determination theory as an analytical lens, we examined how the design and functionality shape nurses' experiences of autonomy, competence, and relatedness, which are three core psychological needs essential for sustaining intrinsic motivation and engagement. Data were collected through semi-structured interviews with 29 nurses and complemented by contextual inquiries. Our findings reveal that while the system streamlines routine tasks and information organization, it simultaneously constrains nurses' ability to exercise professional judgement, manage their workflow, and sustain meaningful social interactions. Rigid triage logic, fragmented system integration, and the loss of subtle communicative cues in text-based consultations challenge autonomy and competence. Moreover, altered work practices affect collegiality and emotional connection with patients. This study contributes to the human-computer interaction discourse by demonstrating how AI-supported healthcare automation interacts with professionals' motivational dynamics and by highlighting design directions to better align AI-supported automated triage systems with users' psychological needs and working conditions.
Emotional regulation strongly impacts the well-being of neurodivergent (ND) people. While studies have explored technologies to support emotion regulation for NDs at a young age, few have taken participatory approaches involving ND adults across the design process to ground design decisions based on their real-life settings. To explore how wearable technologies can support ND adults' emotional regulation, we conducted an interview study with 20 ND supporters who provide intervention to help NDs manage emotional regulation (S1A) and a 14-week participatory workshop with 13 NDs (S1B). S1A findings emphasized the need for "just-in-time" interventions, customizable designs, and appropriate terminology. S1B informed how S1A's insights can be instantiated to a smartwatch application. A 90-day study with four participants (S2) was conducted to evaluate how WellBe, developed from S1 findings, can provide better assistance to approach emotional dysregulation for ND adults. S2 findings show that participants used WellBe in everyday contexts (e.g., routine disruptions, work stress, and interpersonal conflicts), and that longer-term in-the-wild use surfaced needs for personalization, lightweight reflection, and social support. We contribute implications for future design and methodological considerations for ND emotional regulation technologies.
This paper introduces collaborative visual thinking (CVT) as a qualitative research method designed to engage autistic adults in research about complex and abstract topics. CVT pairs a semi-structured interview with the co-construction of a visual artefact on a collaboration canvas, such as a (digital) whiteboard, based on predefined guiding concepts. The method was developed for a study exploring the personal finance practices of 20 autistic adults. Framework analysis of the transcripts from the research sessions and a feedback survey showed that participants experienced CVT as autism friendly due to the overall process and the visual approach. Challenges revolved around clarity of expectations and sensory demands. CVT offers an adaptable, neurodiversity-affirming approach that invites further experimentation across contexts. Research Highlights Introduces Collaborative Visual Thinking (CVT) as a method combining one-to-one interviews with collaborative visual artefact creation for shared sensemaking with autistic adults.Discusses the effectiveness of the method based on a case study with 20 autistic adults.Concludes that CVT may be particularly suited to exploring abstract and interconnected topics, but requires balancing of cognitive and sensory demands.
Disability rights advocates often call for digital content and technologies to be developed using a "born-accessible" approach. There are many potential benefits to using a born-accessible approach, but nowhere in human-computer interaction research or practice literature is the term or approach for "born-accessible" defined. At this point, born-accessible is more of a call to action from the disability community and accessibility researchers rather than an actual methodology to be incorporated into the digital technology research, design, and development lifecycle. The goal of this article is to detail how the Born-Accessible Methodological Framework (BAMF) was defined and developed in three stages: (a) identifying the data that directly informs the conceptual and theoretical underpinnings of the born-accessible approach, (b) the steps required to develop the methodological framework, and then (c) further refining the framework by testing it with two case studies with very different accessibility goals. Through this process, the original four concepts of BAMF were extended into seven concepts. The discussion then reflects upon this process, and it details the resulting BAMF, its benefits, practical implications, underlying assumptions, current limitations, and avenues for future research.
The study of emotion regulation (ER) has emerged as a major field within psychology; in parallel, immersive virtual reality (IVR) is increasingly used to deliver both simulation training and therapeutic interventions. Lately, these fields have begun to merge, with IVR used to train ER skills. In order to realize the potential of combining IVR and ER, this research provides an overview of the field, the contexts in which IVR and its technological affordances have been used, and the empirical outcomes of the reviewed studies. It is the first work to address this issue, highlighting several gaps in the extant literature and presenting an agenda for future research. Results show that IVR has been employed in relatively narrow contexts, primarily focused on therapeutic implementations replicating the real world. While IVR does not always significantly differ from standard treatment, all reviewed studies supported its potential to positively affect participants. Numerous benefits of IVR use are highlighted, including its customizability, controllability, privacy, accessibility, and attractiveness. Furthermore, game-based IVR offers promising future potential for more engaging and innovative user experiences. Finally, a recent increase of publications in the field, combined with more varied uses of IVR and its affordances, indicate a broadening of the field. RESEARCH HIGHLIGHTS The majority of studies included in the scoping review revealed that, in the context of emotion regulation (ER), immersive virtual reality (IVR) was predominantly used to treat psychological and behavioural disorders. Exposure therapy was found to be the most common use context; related to this, the majority of research featured virtual environments which were designed simply as a direct replication, or as simulations of the real-world.While IVR does not always significantly differ from standard treatment, all reviewed studies supported its potential to positively affect participants.Numerous benefits of IVR use were highlighted, including its customisability, controllability, privacy, accessibility, and attractiveness.The results highlight the fact that IVR is proving to be a beneficial tool to facilitate ER, however, notable scope for further developments still remains, particularly in respect to more fruitfully exploiting the affordances of the IVR technology. The recent increase of published works, combined with more varied uses of IVR and its affordances, indicate a broadening of the field.
As Artificial Intelligence (AI) becomes more integrated into production processes, it is essential that AI-based tools are transparent and easy to understand for end users such as production planners and machine operators. Yet, many existing eXplainable AI (XAI) approaches are too complex for this target group. In this paper, we examine which forms of human-AI interaction are preferred in manufacturing settings to support end users in better understanding and interpreting AI-generated recommendations. Based on a design case study, we identify user requirements for a more accessible interface and develop a high-fidelity prototype that is qualitatively evaluated through semi-structured interviews. Results support a clear preference for local explanations, while global ones often cause confusion and lack adaptability. Furthermore, while XAI can increase transparency, upper management sometimes withholds information from shop floor workers to maintain existing power structures. This highlights how XAI can influence workplace dynamics, potentially shifting responsibilities and decision-making authority. These findings underline the need to align AI explanations with both the informational needs of production employees and the strategic concerns of management. Doing so can support more effective and context-sensitive implementation of AI systems in production planning. RESEARCH HIGHLIGHTS Using AI predictions within the production planning process has a great impact on current work practices from production planners and also machine operatorsCurrent XAI approaches mostly lack the ability to provide end users with the explanations they need to make reliable decisionsWhile explanatory approaches can offer more transparency, upper management deliberately chooses not to pass certain information down to the shop floor to avoid a potential shift of power
Human-AI collaboration is considered the most promising way to incorporate AI in the workplace. What remains unexplored are the experiential consequences of this teaming. More specifically, in a team with AI, how humans perceive themselves (self-perception) and how they are perceived by their coworkers (peer perception) in terms of work ownership and job meaningfulness. In a $2 imes 2 imes 2$ vignette study (n=50), participants rated perceptions of ownership, affect, job meaningfulness and satisfaction, and role dynamics across two levels (low/high) of AI proactivity and AI competency as within-subject factors, with point-of-view (self perception/peer perception) as between-subjects. Our results showed that AI with low competency or low proactivity generally improved feelings related to ownership, meaningfulness, satisfaction, and role dynamics, and also increased positive affect while reducing negative affect. However, these effects were often influenced by point-of-view. For instance, low AI proactivity resulted in higher job satisfaction from self-perception rather than peer perception. Based on our findings, we argue that designing AI for the future of work solely around performance metrics may not be adequate. Highly competent and proactive AI-driven systems can have undesirable impacts on perceptions of ownership, job identity, social image and team dynamics, and consequently, job meaningfulness. RESEARCH HIGHLIGHTS As AI becomes increasingly integrated into workplaces, considerable effort is being invested in making AI-driven systems more competent and proactive.While this is often assumed to be desirable, this paper examines whether highly competent and proactive AI may also have unintended social and experiential consequences for workers.Through a vignette-based study, we explore how AI competency and proactivity influence how people perceive themselves and how they are perceived by others in human-AI teams.Based on our findings, we argue that designing AI for the future workplace solely around performance metrics may not be adequate.Highly competent and proactive AI systems can have undesirable impacts on perceptions of ownership, job identity, social image, team dynamics, and consequently, job meaningfulness.As workplace AI becomes more capable and autonomous, the human and social consequences deserve equal consideration alongside efficiency and performance.
Mobile apps are often difficult for older adults to use and frequently fail to meet their needs-an issue that has been well documented in usability studies involving this population. However, there is limited information and few guidelines on how to effectively involve older adults in the mobile app development process to promote the design of user-friendly and accessible interfaces. To address this gap, we conducted a scoping review using Web of Science, Scopus, and the ACM Digital Library, selecting 56 articles that reported user research practices involving older adults to evaluate the usability and user experience of mobile apps. Using quantitative and qualitative analyses, we examined how older users are portrayed as end users, the types of mobile apps that include older adults in their development process, and the research methods applied. Our findings indicate that the literature often reinforces deficit-based narratives, framing older adults primarily in terms of cognitive and physical decline. Accordingly, most studies focused on mHealth apps. We identified four main approaches to user research with older adults: usability testing (22), participatory design (13), follow-up studies (12), and app exploration (9). Based on these results, we advocate for adopting a Universal Design perspective that includes older adults as end users not only for mHealth services but for all types of mobile applications. Recognizing the diversity of senior users is equally important for gathering meaningful insights. We conclude by providing recommendations to improve usability and user experience research practices with mobile apps that involve this population.
Technology is mainly developed by and designed for neurotypical people. Although neurodivergent perspectives are increasingly included in research projects, accessibility within industrial contexts remains insufficiently addressed. This work focuses on (1) work technologies developed in (2) small- and medium-sized enterprises (SMEs), both largely unattended aspects of accessibility research. Our project aims to enhance workplace participation by analyzing situated practices within industry settings. We examine how everyday reasoning processes shape decisions around accessibility, moving beyond explanations based solely on limited awareness or resources. Data collected through questionnaires and interviews with software development companies reveal reasoning patterns that currently hinder opportunities for neurodivergent individuals to structurally be part of the development process. Our data show that SMEs provide unique opportunities in their development processes to cater for individuality within the neurodiversity discussion. However, we identified three mutually reinforcing layers of argumentation that keep neurodiversity considerations at a case-by-case level: (1) accessibility perceived as irrelevant for B2B industries, due to an assumed lack of neurodivergent workforce; (2) the belief that existing individualized approaches already suffice to address neurodiversity requirements; and (3) the perception that repeated structural involvement contradicts these individualized views. To improve actual practice, a deeper understanding of the situated nature of SME work is required, including both its constraints and possibilities as well as underlying societal biases. Addressing all three argumentation layers is essential for increasing initiatives to involve neurodivergent individuals and hence provide workable solutions that recognize structural barriers. RESEARCH HIGHLIGHTS This interview study addresses two under-researched areas in accessibility research: software development processes in small and medium-sized enterprises (SMEs) and the business-to-business (B2B) sector.By investigating everyday decision-making practices and work routines, we sought to identify concrete entry points for structurally embedding neurodivergent perspectives in software development.Rather than finding pathways toward structural inclusion, we uncovered three mutually reinforcing layers of argumentation that explain why B2B SMEs tend to maintain individual, case-by-case accommodation instead.These layers include the underestimation of neurodivergent wokforce in both number and capability, the conflation of accessibility with usability, and a reliance on individual users to selfadvocate for solutions.Collectively, those argumentation layers inhibit the structural anchoring of neurodivergent involvement in development processes.
Despite the rapid integration of artificial intelligence (AI) in professional environments, its implications for employee engagement are not yet well understood. Hence, this study examines how AI technologies impact employee engagement in the IT sector. The study consisted of 28 semi-structured interviews with IT professionals, recruited in Sweden. The responses were recorded, transcribed, and thematically analyzed using the job demands-resource framework. The findings reveal that AI can enhance work engagement by reducing repetitive tasks, supporting learning, and increasing the meaningfulness of work. However, AI also introduces new demands, including cognitive overload, skill relevance uncertainty, and concerns about tool reliability and leadership support. Theoretically, the findings highlight the need to adapt the Job Demands Resources model to account for the dynamic, context-dependent nature of digital technologies, such as AI. This suggests the emergence of a concept we term Digital Work Engagement, a positive and fulfilling user experience of vigor, dedication, and absorption based on the worker's interactions with and relation to technology in the workplace. Practically, the study offers guidance for designers and managers on fostering work engagement by aligning AI development and integration with professional growth, autonomy, and support. RESEARCH HIGHLIGHTS AI simultaneously shapes job resources and demands in IT work.Introduces Digital Work Engagement as a technology-mediated construct.With support and foundations for use, AI enhances work engagement by enabling meaningful work.AI obstruct work engagement through demands of cognitive load, uncertainty, and distrust.
As artificial intelligence systems increasingly participate in emotionally sensitive interactions, artificial empathy can both enhance and undermine users' responses. This study examines how artificial empathy influences perceived usefulness, trust, and intention to use, and whether these effects depend on interaction orientation (emotional vs. functional). We conducted two scenario-based experiments (N = 480). Study 1 establishes baseline relationships between artificial empathy, perceived usefulness, trust, and intention. Study 2 employs a 2 & times; 2 between-subjects design to test moderation by interaction orientation. Results show that artificial empathy increases perceived usefulness, trust, which in turn predict usage intention. These positive effects are found to be significantly weaker in emotional interactions than in functional interactions, which points to the contextual boundary of the empathy paradox that empathic cues may raise authenticity-related concerns and reduce acceptance, especially in emotional settings. These findings highlight the need for context-aware calibration of empathic cues in human-artificial intelligence interaction.
Much HCI research has explored user experience and, in particular, emotional ones, reflected in the increasing interest in well-being and affective health technologies. While much such research has focused on emotional awareness and regulation, less work has explored how to design for emotional appraisal. We report a 2-week diary study informed by the conceptual model of Lazarus's appraisal theory to explore emotional appraisal in the wild by proposing two novel design research methods: the Emotional Appraisal Kit and the Emotional Appraisal Diary. Findings indicate that these tools support this shifting in paradigm by providing a holistic method that considers emotional, physical, and computational elements. Findings also highlight the value of such tools in capturing the emotional process, going beyond emotional recognition to provide a deeper insights into what triggers emotions.
Gesture-based interaction in smart vehicles has been introduced to improve usability, minimize distraction, and enhance road safety. However, existing gestures vary across vehicle brands and are not always intuitive for users. This study examines current trends in gesture control technology and investigates user preferences for in-vehicle gesture interfaces. In particular, it explores the inclusion of American Sign Language (ASL) as a gesture set and analyzes potential gender differences in preferences. A mixed-methods approach was employed: (i) a quantitative closed card sorting study evaluating predefined gestures from commercial vehicle systems and ASL and (ii) a qualitative gesture elicitation study capturing user-defined proposals. Results show that simple one-handed gestures, such as swiping and pointing, were consistently preferred for their versatility and low cognitive demand. ASL gestures were generally perceived as unsuitable for vehicle control due to their complexity and frequent requirement of two-handed execution. Gender analysis revealed no significant differences in gesture preferences. While the relatively small sample size (n = 19 for the elicitation study) limits generalizability, the findings provide design implications for developing intuitive gesture control systems in smart vehicles. They highlight the importance of simplicity, consistency, and integration with other modalities (e.g., voice or touch) to support safer and more effective multimodal interaction.
This paper examines the transition toward Industry 5.0 (I5.0) by foregrounding the role of institutional arrangements in shaping meaningful blue-collar work. Building on prior research, we recontextualize six dimensions of meaningful work-Competence, Autonomy, Embodiment, Relatedness, Beneficence, and Identity-within industrial manufacturing. Moving beyond individual experiences, we identify twelve institutionalized social structures-four normative, three regulative, and five cultural-cognitive-that enable and constrain meaningful work through shared rules, norms, and beliefs. Introducing a multi-level perspective, the study further explores ways in which emerging technologies can reinforce or reconfigure these meso-level arrangements to sustain micro-level experiences of meaningful work and support macro-level transition to I5.0 through socio-technical alignment. Drawing on an abductive analysis of interview data (n = 31) across nine Finnish manufacturing companies, we develop a conceptual framework that supports technology adoption while strengthening work meaningfulness. The findings provide practical guidance for organizations and contribute to advancing scholarly debates on sustainable, human-centered futures of industrial work.
As conversational AI systems proliferate across platforms and use contexts, understanding the structural patterns of human-AI interaction becomes critical for both system design and user experience optimization. We analyzed 1,469,549 conversations from two large-scale datasets (LMSYS-1 M and WildChat) to examine how conversational structures vary across 25 AI models and two deployment platforms. We extracted structural features through automated computational analysis and applied unsupervised clustering and nonparametric statistical tests to identify systematic differences in message length, turn-taking patterns, and conversational balance. Three key findings emerged: (a) deployment context shapes interaction patterns more strongly than model architecture (r = 0.371 vs. r = 0.283), with the same models producing dramatically different conversational structures depending on platform infrastructure, user populations, and task framings; (b) AI models differ substantially in response verbosity, with some generating responses 4.4 times longer than others despite similar capabilities; (c) four distinct conversation types emerged across datasets (technical assistance, general Q&A, intensive collaboration, and quick lookups) with 97.5% consisting of single-exchange interactions rather than multiturn dialogue. These findings challenge assumptions about human-AI conversation as dialogic exchange and demonstrate that deployment context, user populations, and platform affordances fundamentally shape interaction patterns independent of technical capabilities. We discuss implications for conversational AI design, evaluation practices, and theoretical frameworks for understanding human-AI communication.
Predicting user attention within complex interfaces presents a significant challenge for designers. Traditional heuristic approaches, such as visual hierarchy, have demonstrated limited empirical validity. We present computational salience modeling as a potential alternative to a classic heuristic approach. In doing so, we summarize major advances in salience modeling and explore what is needed for this computational approach to become an effective tool. Previous research has demonstrated the robustness of salience model predictions within the context of interface design and task manipulations. Salience models have reliably predicted the deployment of overt attention across diverse interface contexts, including webpages and mobile displays. Notably, they remain predictive under both free-viewing and goal-directed search tasks, and across different levels of visual clutter. Higher levels of visual salience consistently improve search efficiency. Recent models show enhanced predictive performance by integrating computational salience with empirically derived spatial convention maps reflecting learned experiences. Computational salience modeling can provide designers with a useful, data-driven method for predicting attentional guidance. It offers objective identification of visually salient elements and provides insight into improved search efficiency.
