Generative artificial intelligence is increasingly embedded in architectural practice and education, yet its role often remains confined to image production or optimization tasks. This study situates generative AI within a broader design ecology. It examines how structured human–AI interaction can support environmentally oriented architectural thinking in design education. The article presents an international design workshop as a research setting in which architecture students engaged with AI through a multi-agent workflow. This workflow combined large language models, diffusion-based image generation, 2D–3D translation tools, parametric modeling, and clay-based 3D printing. Central to the methodology is the concept of prompt choreographies. These are deliberate dialogs between human and AI agents, based on a language of prompts and AI-generated outcomes. Through this process, the design concept moves toward a final architectural proposal. The workshop addressed complex ecological challenges emerging from interactions among Earth’s spheres. These were conceived as environmental interfaces defined by behavioral continuity rather than typological form. Using qualitative, design-based evaluation criteria focused on environmental, spatial, and material aspects, the study identifies recurring patterns of human–AI collaboration. The findings indicate that generative AI supports architectural ideation most effectively when embedded in structured workflows that emphasize curatorial decision-making and reduce generative overproduction. While limited to a workshop-based educational context, the research offers transferable methodological insights for architectural pedagogy and conceptual practice. It proposes a process-oriented framework for designing with generative AI and outlines an emerging form of architectural literacy and multi-agent collaboration that warrants further empirical validation.

Learning Health Systems provide a glide path to safe landing for AI in health.

Comparison of VADER and TextBlob labeling for sentiment analysis using machine learning and deep learning models: A study on generative AI user experience.

A qualitative shift in AI capabilities: A “bitter lesson” for human-AI interaction research?

The use of Artificial Intelligence (AI) in learning is expanding globally; however, the full potential of AI tools in the Open and Distance Learning (ODL) context, particularly at the Institute of Adult Education (IAE), remains underexplored. This study examined the IAE ODL students’ perspectives on the use of AI tools in learning. Specifically, it investigated ODL students’ familiarity with AI, AI preferences and use in learning, and perspectives on AI tool use in ODL. The study employed a mixed-methods approach, utilising a convergent parallel design to collect data from 93 second- and third-year ODL students at the Dar es Salaam and Morogoro Campuses. The findings revealed that 94.7% of students were familiar with AI, mainly after beginning their studies; 87% used ChatGPT for learning, and 57% used AI to answer their questions. In addition, 98% of students argued that the utilisation of AI in ODL is inevitable, citing its role in enhancing self-learning, improving access to learning materials, and saving time. Based on the findings, the study suggests that enhanced access to and awareness of diverse AI tools may help maximise their potential benefits in learning. The study also calls for academic integrity, ethical use, peer learning, and human-AI interaction among ODL students and institutions for the effective utilisation of AI in ODL.

This paper presents the design of a Brazilian robot, named Zequinha, for cultural and educational purposes in light of Human-Centered Artificial Intelligence challenges. Zequinha's development is a blend of art, robotics, and AI, evolving from MIDI-programmed animatronics to an autonomous entity integrating multiple local AIs. This shift to local processing inherently enhances privacy and governance, minimizing reliance on external APIs and enabling offline operability. The project's human-centered design approach is evident in its iterative methodology and its collaborative development with children. Zequinha promotes human wellbeing by enriching cultural mediation, engaging diverse audiences, and demonstrating potential in health and education. Moreover, the focus on local AI fosters responsible design and meaningful human-AI interaction, aiming to create a charismatic, safe, and useful robotic mediator.

The understanding of gestures and the synthesis of choreography can be viewed as two distinct sides of the human-AI interaction problem, which cannot be viewed as complementary and must be addressed through joint modeling of perception, synthesis, and real-time interaction. An interactive multimodal neural architecture consisting of spatial-temporal gesture encoding, latent motion representation learning, and style-conditioned choreography synthesis is proposed to facilitate end-to-end transfer of human movement from sense to expressive synthesized movement. The semantic consistency constraints in joint optimization will be used to ensure consistency between the perceived gesture intent and the synthesized choreography, while an edge cloud deployment approach will be utilized to facilitate interactive latency and energy-efficient execution. The experimental evaluation on benchmark datasets and live co-creative applications demonstrate high recognition accuracy, smooth and diverse motion synthesis, and successful semantic agreement and consistency in co-creating real-time settings. The formal user study also reveals high levels of perceptual realism, sense of expression, usability, and creative satisfaction, which verifies the framework as an excellent collaborative partner and not a passive generative tool. Managerial analysis Networks have lower production costs, scalable deployment opportunities, and therapeutic engagement of benefits in the areas of creative media, rehabilitation, and social robotics. The findings place gesture-based creative AI as a promising foundation of embodied intelligent interaction, and future research directions include the integration of emotion in creative choreography synthesis, adaptive reinforcement learning co-creation, and extreme low-latency edge synthesis

This study addresses the need for responsible AI adoption in higher education by developing a human-centred ethical extension of the UTAUT2 model. It integrates two new constructs; AI fairness and human autonomy support and three ethical moderators: ethical risk awareness, perceived algorithm bias and user autonomy concern. To validate the framework, an empirical investigation was conducted with 400 respondents using a structured questionnaire, with data analyzed via regression. All sixteen hypotheses were supported. The model demonstrated strong predictive power, explaining 72.2% of the variance in behavioural intention and 69.1% in use behaviour. The results provide meaningful insights into how ethical perceptions influence adoption. Ultimately, the framework offers practical guidance for policymakers, educators and developers to ensure fair, trustworthy and human-centric AI integration in learning environments.

AI technologies have been widely explored in the architectural design process over the last decade. This paper addresses the limitations of the existing reviews on AI applications in the architectural design process, including a lack of focus on the design process, limited coverage of AI technologies, and under-exploration of human-AI interaction environments. The paper systematically reviews and comparatively analyses 63 articles filtered through 1138 publications from 8 databases. The findings include comparative analysis charts and tables of the different AI-enhanced design processes, the human-AI interaction environment, as well as their evaluation. They highlight the expectation that AI will function as a conversational design partner. Lastly, the paper presents a novel framework for the AI-enhanced conversational architectural design process.

High-entropy alloys (HEAs) are emerging as next-generation structural materials due to their outstanding mechanical and functional properties. However, their vast compositional and configurational complexity poses major challenges for conventional trial-and-error approaches and ab initio simulations, which struggle with high computational costs and limited predictive accuracy. Existing machine learning approaches, while promising, remain constrained by data scarcity, limited interpretability, and the lack of effective human-AI interaction. To address these limitations, we introduce an integrated human-computer interactive HEA design platform that incorporates emotional feedback. By combining natural language processing (T5 model), machine learning (XGBoost), and multi-objective optimization (NSGA-II), the platform establishes a closed-loop "perception-decision-optimization" workflow. Real-time emotion recognition dynamically adjusts optimization weights, enabling efficient human-AI collaboration. The model achieves high accuracy in predicting yield strength and Young's modulus, with SHAP analysis revealing the underlying physical mechanisms. Emotion-driven optimization guides Pareto front convergence, with results showing <1.4% deviation from experimental values. This high degree of accuracy underscores the efficacy of integrating affective feedback into the optimization loop, enabling a more responsive and user-aligned design process that effectively bridges subjective expert preferences with quantitative multi-objective optimization. The multiscale modeling further validates the platform's reliability for complex dual-phase alloys. This work establishes a novel paradigm for interpretable and efficient AI-driven material design, highlighting the transformative potential of integrating artificial intelligence with expert knowledge.

State-wise linear modulation (SLim): A novel approach for steering large language models.

The integration of Generative AI (GenAI) into educational settings necessitates a paradigm shift for collaborative learning, as established frameworks remain undertheorized for non-human agents. Extending the empirically grounded Collaborative Idea Construction (CIC) model and its stable peer roles (MKP, PA, DP), this paper proposes the AI- Enhanced CIC (AI-CIC) framework. GenAI is conceptualized as a novel 'Peer Apprentice' rather than a simple tool. This intervention reshapes human social architecture, transforming the Peer Assistant into an 'AI Validator' and elevating the More Knowledgeable Peer to a 'Metacognitive Facilitator.' Consequently, the collaborative workflow is hypothesized to become more linear. The AI-CIC model offers a theory-driven framework with testable propositions regarding emergent socio-technical roles and workflow shifts. It provides a foundation for future empirical research and informs pedagogical strategies for facilitating critical human-AI interaction.

Artificial Intelligence was originally conceived as a tool for human ease, intended to enhance productivity and streamline complex tasks. However, recent trends suggest a shift from augmentation to a total "cognitive offloading," where humans increasingly outsource critical thinking and memory to AI. This review paper explores the hypothesis that this dependency is contributing to a reversal of the Flynn Effect. After reviewing neuroimaging data from fMRI and EEG studies, the paper identifies significant alterations in neuroplasticity and neural activity. A primary concern is the observed decline in hippocampal volume among individuals who over-rely on AI for routine and complex cognitive tasks. Because the brain follows a "use it or lose it" paradigm, bypassing "desirable difficulties" during information processing leads to a lack of deep memory encoding and weakened retention skills. This phenomenon, often termed "Digital Amnesia," suggests that our neural architecture is physically adapting to a state of passive retrieval rather than active synthesis. The paper concludes that while AI offers immense efficiency, we must develop a balanced Human-AI interaction model. By treating AI as a collaborative partner rather than a cognitive substitute, we can leverage technological speed while preserving the biological integrity and intellectual capacity of the human brain.

The growing integration of generative Artificial Intelligence (AI) in workplaces has transformed the way professionals think, learn and perform tasks. While AI enhances efficiency and accessibility, increasing dependence on it may influence employee’s cognitive engagement and their psychological connection with work. The present study aims to examine the relationship of Generative AI Dependency with Work Alienation and Need for Cognition among working professionals. The study is grounded in the assumption that greater reliance on AI may be associated with increased alienation from work and reduced inclination towards cognitive effort. It is hypothesized that Hypothesis 1: there will be a significant positive relationship between Generative AI Dependency and Work Alienation and Hypothesis 2: there will be a significant negative relationship between Generative AI Dependency and Need for Cognition. A quantitative correlational design will be employed and data will be collected using standardized measures: the Generative AI Dependency Scale (Goh, Hartanto &amp; Majeed, 2025), the Work Alienation Scale (Nair &amp; Vohra, 2010) and the Need for Cognition Scale (Cacioppo, Petty &amp; Kao, 1984), each scale consisting of 11, 8 and 18 items respectively. The questionnaires will be administered through Google Forms to a sample of working professionals. Pearson correlation will be used to analyze the relationship among variables. The study seeks to contribute to emerging literature on the psychological implications of AI usage by exploring whether reliance on generative AI is associated with greater feelings of detachment from work and reduced inclination towards cognitive effort. Findings are expected to inform balanced human-AI interaction strategies that preserve critical thinking and employee’s sense of meaning within organizational settings.

ABSTRACT This paper investigates prosopopoeia, the rhetorical act of giving voice or persona to nonhuman entities, as a framework for addressing the communicative anxieties that emerge in human–AI interaction. Drawing on Gregory Bateson’s theory of play and metacommunication, the paper develops prosopopoeia as a third mode between naïve anthropomorphism and instrumental tool use. Unlike anthropomorphism, which attributes human qualities to AI and can erode critical human agency, deliberate prosopopoeia treats the machine as a performance to be managed rather than a subject to be believed. The argument is illustrated with empirical material from a university course in which students’ struggles with chatbot interactions revealed a lack of stable metacommunicative frames and loss of agency, and failed to resolve their interpretive uncertainty. By stressing the metacommunicative aspect of human–AI interaction as conscious performance, this paper outlines prosopopoeic frameworks as a critical competency for engaging with AI systems. The study contributes to human–machine communication by offering a theoretical framework for developing new forms of AI literacy. Prosopopoeia provides a method for engaging meaningfully with entities whose communicative and ontological status is fundamentally ambiguous, emphasizing performative awareness over anthropomorphic projection.

Artificial intelligence (AI) chatbots have become prominent tools in health care to enhance health knowledge and promote healthy behaviors across diverse populations. However, factors influencing the perception of AI chatbots and human-AI interaction are largely unknown. This study aimed to identify interaction characteristics associated with the perception of an AI chatbot identity as a human versus an artificial agent, adjusting for sociodemographic status and previous chatbot use in a diverse sample of women. This study was a secondary analysis of data from the HeartBot trial in women aged 25 years or older who were recruited through social media from October 2023 to January 2024. The original goal of the HeartBot trial was to evaluate the change in awareness and knowledge of heart attack after interacting with a fully automated AI HeartBot chatbot. All participants interacted with HeartBot once. At the beginning of the conversation, the chatbot introduced itself as HeartBot. However, it did not explicitly indicate that participants would be interacting with an AI system. The perceived chatbot identity (human vs artificial agent), conversation length with HeartBot, message humanness, message effectiveness, and attitude toward AI were measured at the postchatbot survey. Multivariable logistic regression was conducted to explore factors predicting women's perception of a chatbot's identity as a human, adjusting for age, race or ethnicity, education, previous AI chatbot use, message humanness, message effectiveness, and attitude toward AI. Among 92 women (mean age 45.9, SD 11.9; range 26-70 y), the chatbot identity was correctly identified by two-thirds (n=61, 66%) of the sample, while one-third (n=31, 34%) misidentified the chatbot as a human. Over half (n=53, 58%) had previous AI chatbot experience. On average, participants interacted with the HeartBot for 13.0 (SD 7.8) minutes and entered 82.5 (SD 61.9) words. In multivariable analysis, only message humanness was significantly associated with the perception of chatbot identity as a human compared with an artificial agent (adjusted odds ratio 2.37, 95% CI 1.26-4.48; P=.007). To the best of our knowledge, this is the first study to explicitly ask participants whether they perceive an interaction as human or from a chatbot (HeartBot) in the health care field. This study's findings (role and importance of message humanness) provide new insights into designing chatbots. However, the current evidence remains preliminary. Future research is warranted to understand the relationship between chatbot identity, message humanness, and health outcomes in a larger-scale study.

ABSTRACT Evaluating creative media based on the type of creator (AI vs. human) offers fertile ground for theorising and deepening our understanding of human-AI interaction. Traditionally, creative works have been evaluated by the artistic value attributed to their creators. In media creation, AI can be seen as a valuable tool that enhances creators’ productivity and compensates for gaps in their abilities. However, AI can also be viewed as a mechanistic approach to media production, allowing large studios to create content with minimal artistic input. Drawing from prior literature, we theorise how the use of generative AI may impact the perceived value of art and the mechanisms by which this value is shaped. Building on Berlyne’s model of aesthetic experience, we show that the use of generative AI significantly influences appraisers’ perceptions of an artwork's complexity, thereby affecting its value. Interestingly, the type of creator did not affect the emotional response to the art. We discuss the implications of these findings.

Purpose To ensure the effective utilization of information resources by users in the era of artificial intelligence, it is crucial to explore the factors influencing user information adoption behavior and its configurational pathways within human–AI interaction contexts, which is the aim of this study. Design/methodology/approach This study focuses on users of AIGC platforms and employs the Elaboration Likelihood Model (ELM) as a theoretical foundation. Data analysis is conducted using Structural Equation Modeling (SEM) and fuzzy-set Qualitative Comparative Analysis (fsQCA). Findings The SEM results indicate that, with the exception of technological characteristics, all other factors positively influence user information adoption behavior. The fsQCA identifies four distinct configurations that contribute to information adoption behavior. Originality/value The findings suggest that AIGC platforms should enhance user information adoption by optimizing interaction systems, ensuring information quality, simplifying operational processes, and integrating emotional design.

The performance of contrast-enhanced computed tomography (CECT) in staging clear cell renal cell carcinomas (ccRCCs) and assessing tumor aggressiveness remains limited by heterogeneous and poor sensitivity. This study aims to design and validate a human-AI interactive network, Kidney Tumor Staging Network (KtSNet), which leverages weakly supervised learning for real-time, efficient detection of high-grade aggressive ccRCC (HGRCC) using CECT. A total of 1,092 patients with ccRCC were enrolled across five cohort datasets (training/internal testing/external testing, n=611/153/328). To achieve precise pre-surgical detection of HGRCC on CT imaging, we pretrained a self-supervised foundation model (SSFM) using a large cross-modal dataset (n=40000) for image restoration-based transfer learning. To develop human-AI interactive capabilities, we trained KtSNet by integrating SSFM with weakly supervised learning, enabling real-time determination of HGRCC on CT imaging through human-AI interaction. In the internal test cohort comprising 153 patients, KtSNet demonstrated significantly higher Area Under the Curve (AUC) values for both ROC and PR curves (ROC-AUC=0.76; PR-AUC=0.29, F1 max: 0.441) compared to B2Net (ROC-AUC=0.68, p=0.040; PR-AUC=0.22, F1 max: 0.366), RML-XGB (ROC-AUC=0.53, p<0.001; PR-AUC=0.14, F1 max: 0.264), and Likert scoring (ROC-AUC values of 0.57, 0.58, and 0.70 for the three readers) in staging HGRCC. In the external validation cohort, KtSNet maintained superior AUCs on both ROC and PR curves (ROC-AUC=0.85; PR-AUC=0.42, F1 max: 0.529) compared to B2Net (ROC-AUC=0.74, p=0.002; PR-AUC=0.21, F1 max: 0.328) and exhibited significantly higher AUCs than RML-XGB (ROC-AUC=0.63, p=0.002; PR-AUC=0.23, F1 max: 0.359). The weakly human-supervised KtSNet may serve as a promising opportunity for real-time determination of HGRCC using CT imaging.

Who is responsible for self-AI or others-AI collaboration? The effect of power and task outcome in responsibility attribution.