Human-centered Artificial Intelligence Research Articles

Designing human-centered learning analytics and artificial intelligence in education solutions: a systematic literature review

ABSTRACT The recent advances in educational technology enabled the development of solutions that collect and analyse data from learning scenarios to inform the decision-making processes. Research fields like Learning Analytics (LA) and Artificial Intelligence (AI) aim at supporting teaching and learning by using such solutions. However, their adoption in authentic settings is still limited, among other reasons, derived from ignoring the stakeholders' needs, a lack of pedagogical contextualisation, and a low trust in new technologies. Thus, the research fields of Human-Centered LA (HCLA) and Human-Centered AI (HCAI) recently emerged, aiming to understand the active involvement of stakeholders in the creation of such proposals. This paper presents a systematic literature review of 47 empirical research studies on the topic. The results show that more than two-thirds of the papers involve stakeholders in the design of the solutions, while fewer papers involved them during the ideation and prototyping, and the majority do not report any evaluation. Interestingly, while multiple techniques were used to collect data (mainly interviews, focus groups and workshops), few papers explicitly mentioned the adoption of existing HC design guidelines. Further evidence is needed to show the real impact of HCLA/HCAI approaches (e.g., in terms of user satisfaction and adoption).

Verified localization and pharmacognosy of herbal medicinal plants in a combined network framework

Pharmacognosy from medicinal plants involves the scientific domain of medicinal compounding based on their medicinal properties. Accurate identification of medicinal plants is crucial, especially by examining their leaves. Choosing the wrong plant species for medicinal preparations can have adverse side effects. This study presents a Human-Centered Artificial Intelligence approach for medicinal plant identification, combining a YOLOv7-based Leaf Localizer with a leaf Class Verifier based on DenseNet through a Confidence Score Analyser algorithm. The Confidence Score Analyser ensures reliability by evaluating predicted categories against predefined thresholds, and the ensemble technique through majority voting enhances robustness. An average performance gain of 25.66% sensitivity is observed when comparing the YOLO object detection model with 77.45% precision to the YOLO integrated with the class verifier model with 97.33% precision. Consistent sensitivities are achieved through the ensemble technique, showcasing robustness across diverse scenarios. The final step incorporates automated textual and audio pharmacognosy information about the identified medicinal plant properties and their utility. Real-time applicability as a smart phone application makes this approach invaluable for medicinal plant collectors and experts.

Build Your Own Robot Friend: An Open-Source Learning Module for Accessible and Engaging AI Education

As artificial intelligence (AI) is playing an increasingly important role in our society and global economy, AI education and literacy have become necessary components in college and K-12 education to prepare students for an AI-powered society. However, current AI curricula have not yet been made accessible and engaging enough for students and schools from all socio-economic backgrounds with different educational goals. In this work, we developed an open-source learning module for college and high school students, which allows students to build their own robot companion from the ground up. This open platform can be used to provide hands-on experience and introductory knowledge about various aspects of AI, including robotics, machine learning (ML), software engineering, and mechanical engineering. Because of the social and personal nature of a socially assistive robot companion, this module also puts a special emphasis on human-centered AI, enabling students to develop a better understanding of human-AI interaction and AI ethics through hands-on learning activities. With open-source documentation, assembling manuals and affordable materials, students from different socio-economic backgrounds can personalize their learning experience based on their individual educational goals. To evaluate the student-perceived quality of our module, we conducted a usability testing workshop with 15 college students recruited from a minority-serving institution. Our results indicate that our AI module is effective, easy-to-follow, and engaging, and it increases student interest in studying AI/ML and robotics in the future. We hope that this work will contribute toward accessible and engaging AI education in human-AI interaction for college and high school students.

Leveraging code-free deep learning for pill recognition in clinical settings: A multicenter, real-world study of performance across multiple platforms

BackgroundPreventable patient harm, particularly medication errors, represent significant challenges in healthcare settings. Dispensing the wrong medication is often associated with mix-up of lookalike and soundalike drugs in high workload environments. Replacing manual dispensing with automated unit dose and medication dispensing systems to reduce medication errors is not always feasible in clinical facilities experiencing high patient turn-around or frequent dose changes. Artificial intelligence (AI) based pill recognition tools and smartphone applications could potentially aid healthcare workers in identifying pills in situations where more advanced dispensing systems are not implemented. ObjectiveMost of the published research on pill recognition focuses on theoretical aspects of model development using traditional coding and deep learning methods. The use of code-free deep learning (CFDL) as a practical alternative for accessible model development, and implementation of such models in tools intended to aid decision making in clinical settings, remains largely unexplored. In this study, we sought to address this gap in existing literature by investigating whether CFDL is a viable approach for developing pill recognition models using a custom dataset, followed by a thorough evaluation of the model across various deployment scenarios, and in multicenter clinical settings. Furthermore, we aimed to highlight challenges and propose solutions to achieve optimal performance and real-world applicability of pill recognition models, including when deployed on smartphone applications. MethodsA pill recognition model was developed utilizing Microsoft Azure Custom Vision platform and a large custom training dataset of 26,880 images captured from the top 30 most dispensed solid oral dosage forms (SODFs) at the three participating hospitals. A comprehensive internal and external testing strategy was devised, model's performance was investigated through the online API, and offline using exported TensorFlow Lite model running on a Windows PC and on Android, using a tailor-made testing smartphone application. Additionally, model's calibration, degree of reliance on color features and device dependency was thoroughly evaluated. Real-world performance was assessed using images captured by hospital pharmacists at three participating clinical centers. ResultsThe pill recognition model showed high performance in Microsoft Azure Custom Vision platform with 98.7 % precision, 95.1 % recall, and 98.2 % mean average precision (mAP), with thresholds set to 50 %. During internal testing utilizing the online API, the model reached 93.7 % precision, 88.96 % recall, 90.81 % F1-score and 87.35 % mAP. Testing the offline TensorFlow Lite model on Windows PC showed a slight performance reduction, with 91.16 % precision, 83.82 % recall, 86.18 % F1-score and 82.55 % mAP. Performance of the model running offline on the Android application was further reduced to 86.50 % precision, 75.00 % recall, 77.83 % F1-score and 69.24 % mAP. During external clinical testing through the online API an overall precision of 83.10 %, recall of 71.39 %, and F1-score of 75.76 % was achieved. ConclusionOur study demonstrates that using a CFDL approach is a feasible and cost-effective method for developing AI-based pill recognition systems. Despite the limitations encountered, our model performed well, particularly when accessed through the online API. The use of CFDL facilitates interdisciplinary collaboration, resulting in human-centered AI models with enhanced real-world applicability. We suggest that rather than striving to build a universally applicable pill recognition system, models should be tailored to the medications in a regional formulary or needs of a specific clinic, which can in turn lead to improved performance in real-world deployment in these locations. Parallel to focusing on model development, it is crucial to employ a human centered approach by training the end users on how to properly interact with the AI based system to maximize benefits. Future research is needed on refining pill recognition models for broader adaptability. This includes investigating image pre-processing and optimization techniques to enhance offline performance and operation on handheld devices. Moreover, future studies should explore methods to overcome limitations of CFDL development to enhance the robustness of models and reduce overfitting. Collaborative efforts between researchers in this domain and sharing of best practices are vital to improve pill recognition systems, ultimately enhancing patient safety and healthcare outcomes.

TachAId-An interactive tool supporting the design of human-centered AI solutions.

In an era where Artificial Intelligence (AI) integration into business processes is crucial for maintaining competitiveness, there is a growing need for structured guidance on designing AI solutions that align with human needs. To this end, we present "technical assistance concerning human-centered AI development" (tachAId), an interactive advisory tool which comprehensively guides AI developers and decision makers in navigating the machine learning lifecycle with a focus on human-centered design. tachAId motivates and presents concrete technical advice to ensure human-centeredness across the phases of AI development. The tool's effectiveness is evaluated through a catalog of criteria for human-centered AI in the form of relevant challenges and goals, derived from existing methodologies and guidelines. Lastly, tachAId and one other comparable advisory tool were examined to determine their adherence to these criteria in order to provide an overview of the human-centered aspects covered by these tools and to allow interested parties to quickly assess whether the tools meet their needs.

The Ethical Challenges of Educational Artificial Intelligence and Coping Measures: A Discussion in the Context of the 2024 World Digital Education Conference

Artificial intelligence (AI), as the core technology of the fourth industrial revolution, has been widely deployed in many areas, bringing tremendous changes to human society. At the same time, AI has also instigated a variety of ethical issues regarding basic human rights, social order, private safety, and more. In order to maintain a balance between technological development and the ethics of AI, governments of various countries and international organizations are working to develop AI regulations and ethical norms. A forum themed “Artificial Intelligence and Digital Ethics” was held as a side event during the 2024 World Digital Education Conference (WDEC), showcasing the Chinese government’s adherence to the ethical notion of “human-centered AI” and the principle of “digital for good” in using AI in digital education. The forum emphasized the importance of establishing the ethics of educational AI for circumventing relevant ethical risks and creating healthy environments for the digital transformation of education. Based on the forum’s theme, this article seeks to set forth the necessity of formulating a code of ethical norms for educational AI and to explore pathways to building an ethical framework in this regard in order to provide insights into the rational application of AI in education and promote the sustainable development of digital education.

AI Lifecycle Zero-Touch Orchestration within the Edge-to-Cloud Continuum for Industry 5.0

The advancements in human-centered artificial intelligence (HCAI) systems for Industry 5.0 is a new phase of industrialization that places the worker at the center of the production process and uses new technologies to increase prosperity beyond jobs and growth. HCAI presents new objectives that were unreachable by either humans or machines alone, but this also comes with a new set of challenges. Our proposed method accomplishes this through the knowlEdge architecture, which enables human operators to implement AI solutions using a zero-touch framework. It relies on containerized AI model training and execution, supported by a robust data pipeline and rounded off with human feedback and evaluation interfaces. The result is a platform built from a number of components, spanning all major areas of the AI lifecycle. We outline both the architectural concepts and implementation guidelines and explain how they advance HCAI systems and Industry 5.0. In this article, we address the problems we encountered while implementing the ideas within the edge-to-cloud continuum. Further improvements to our approach may enhance the use of AI in Industry 5.0 and strengthen trust in AI systems.

Artificial intelligence (AI)-it's the end of the tox as we know it (and I feel fine).

The rapid progress of AI impacts diverse scientific disciplines, including toxicology, and has the potential to transform chemical safety evaluation. Toxicology has evolved from an empirical science focused on observing apical outcomes of chemical exposure, to a data-rich field ripe for AI integration. The volume, variety and velocity of toxicological data from legacy studies, literature, high-throughput assays, sensor technologies and omics approaches create opportunities but also complexities that AI can help address. In particular, machine learning is well suited to handle and integrate large, heterogeneous datasets that are both structured and unstructured-a key challenge in modern toxicology. AI methods like deep neural networks, large language models, and natural language processing have successfully predicted toxicity endpoints, analyzed high-throughput data, extracted facts from literature, and generated synthetic data. Beyond automating data capture, analysis, and prediction, AI techniques show promise for accelerating quantitative risk assessment by providing probabilistic outputs to capture uncertainties. AI also enables explanation methods to unravel mechanisms and increase trust in modeled predictions. However, issues like model interpretability, data biases, and transparency currently limit regulatory endorsement of AI. Multidisciplinary collaboration is needed to ensure development of interpretable, robust, and human-centered AI systems. Rather than just automating human tasks at scale, transformative AI can catalyze innovation in how evidence is gathered, data are generated, hypotheses are formed and tested, and tasks are performed to usher new paradigms in chemical safety assessment. Used judiciously, AI has immense potential to advance toxicology into a more predictive, mechanism-based, and evidence-integrated scientific discipline to better safeguard human and environmental wellbeing across diverse populations.

The Impact of Artificial Intelligence on Ghanaian Health Worker Training: Opportunities, Challenges, and Ethical Considerations

Artificial intelligence (AI) integration within Ghana’s health workforce training programs shows immense potential to enhance quality, consistency and personalization if thoughtfully implemented. This mixed methods SWOT analysis evaluates major strengths like expanded accessibility, weaknesses around connectivity barriers and literacy, opportunities in simulation training, and threats of job automation requiring ethical change management. Ghanaian case studies and policy documents inform targeted opportunities to employ context-appropriate design, creative financing partnerships and participatory principles upholding equity. Risks of algorithmic bias and data exploitation spotlight needs for transparent auditing, consent safeguards and democratized oversight. Recommended actions integrate precedents in change leadership, human-centered AI adoption and African innovation prioritization to pragmatically advance both cutting edge advancement and human dignity. With Ghana pioneering amongst first nations globally to release a national AI strategy, demonstration of such balanced governance can model pathways for technological innovation upholding empowerment across sectors and regions. This multidimensional analytical evaluation contributes uniquely inclusive, grounded perspectives to advance conceptual clarity and practical guidance around AI assimilation - directing focus toward preparatory investments, policy formulation, partnership building and design choices required to ensure gains outpacing grievances along ongoing modernization trajectories.

Toward human-centered AI management: Methodological challenges and future directions

As algorithms powered by Artificial Intelligence (AI) are increasingly involved in the management of organizations, it becomes imperative to conduct human-centered AI management research and understand people's feelings and behaviors when machines gain power over humans. The two mainstream methods – vignette studies and case studies – reveal important but inconsistent insights. Here we discuss the respective limitations of vignette studies (affective forecasting errors, biased media coverage, and question substitution) and case studies (social desirability biases and lack of random assignment and control conditions), which may lead them to overrate negative and positive reactions to AI management, respectively. We further discuss the advantages of a third method for mitigating these limitations: field experiments on crowdsourced marketplaces. A proof-of-concept study on Amazon Mechanical Turk (Mturk; as a world-leading crowdsourcing platform) showed unique human reactions to AI management, which were not perfectly aligned with those in vignette or case studies. Participants (N = 504) did not differ significantly under AI versus human management, in terms of performance, intrinsic motivation, fairness perception, and commitment. We suggest that crowdsourced marketplaces can go beyond human research subject pools and become models of AI-managed workplaces, facilitating timely behavioral research and robust predictions on human-centered work designs and organizations.

Responsible Artificial Intelligence and Journal Publishing

The aim of this opinion piece is to examine the responsible use of artificial intelligence (AI) in relation to academic journal publishing. The work discusses approaches to AI with particular attention to recent developments with generative AI. Consensus is noted around eight normative themes for principles for responsible AI and their associated risks. A framework from Shneiderman (2022) for human-centered AI is employed to consider journal publishing practices that can address the principles of responsible AI at different levels. The resultant AI principled governance matrix (AI-PGM) for journal publishing shows how countermeasures for risks can be employed at the levels of the author-researcher team, the organization, the industry, and by government regulation. The AI-PGM allows a structured approach to responsible AI and may be modified as developments with AI unfold. It shows how the whole publishing ecosystem should be considered when looking at the responsible use of AI—not just journal policy itself.

Letter to the Editor: "How Can Biomedical Engineers Help Empower Individuals With Intellectual Disabilities? The Potential Benefits and Challenges of AI Technologies to Support Inclusivity and Transform Lives".

The rapid advancement of Artificial Intelligence (AI) is transforming healthcare and daily life, offering great opportunities but also posing ethical and societal challenges. To ensure AI benefits all individuals, including those with intellectual disabilities, the focus should be on adaptive technology that can adapt to the unique needs of the user. Biomedical engineers have an interdisciplinary background that helps them to lead multidisciplinary teams in the development of human-centered AI solutions. These solutions can personalize learning, enhance communication, and improve accessibility for individuals with intellectual disabilities. Furthermore, AI can aid in healthcare research, diagnostics, and therapy. The ethical use of AI in healthcare and the collaboration of AI with human expertise must be emphasized. Public funding for inclusive research is encouraged, promoting equity and economic growth while empowering those with intellectual disabilities in society.

Designing Human-Centered AI to Prevent Medication Dispensing Errors: Focus Group Study With Pharmacists.

Medication errors, including dispensing errors, represent a substantial worldwide health risk with significant implications in terms of morbidity, mortality, and financial costs. Although pharmacists use methods like barcode scanning and double-checking for dispensing verification, these measures exhibit limitations. The application of artificial intelligence (AI) in pharmacy verification emerges as a potential solution, offering precision, rapid data analysis, and the ability to recognize medications through computer vision. For AI to be embraced, it must be designed with the end user in mind, fostering trust, clear communication, and seamless collaboration between AI and pharmacists. This study aimed to gather pharmacists' feedback in a focus group setting to help inform the initial design of the user interface and iterative designs of the AI prototype. A multidisciplinary research team engaged pharmacists in a 3-stage process to develop a human-centered AI system for medication dispensing verification. To design the AI model, we used a Bayesian neural network that predicts the dispensed pills' National Drug Code (NDC). Discussion scripts regarding how to design the system and feedback in focus groups were collected through audio recordings and professionally transcribed, followed by a content analysis guided by the Systems Engineering Initiative for Patient Safety and Human-Machine Teaming theoretical frameworks. A total of 8 pharmacists participated in 3 rounds of focus groups to identify current challenges in medication dispensing verification, brainstorm solutions, and provide feedback on our AI prototype. Participants considered several teaming scenarios, generally favoring a hybrid teaming model where the AI assists in the verification process and a pharmacist intervenes based on medication risk level and the AI's confidence level. Pharmacists highlighted the need for improving the interpretability of AI systems, such as adding stepwise checkmarks, probability scores, and details about drugs the AI model frequently confuses with the target drug. Pharmacists emphasized the need for simplicity and accessibility. They favored displaying only essential information to prevent overwhelming users with excessive data. Specific design features, such as juxtaposing pill images with their packaging for quick comparisons, were requested. Pharmacists preferred accept, reject, or unsure options. The final prototype interface included (1) checkmarks to compare pill characteristics between the AI-predicted NDC and the prescription's expected NDC, (2) a histogram showing predicted probabilities for the AI-identified NDC, (3) an image of an AI-provided "confused" pill, and (4) an NDC match status (ie, match, unmatched, or unsure). In partnership with pharmacists, we developed a human-centered AI prototype designed to enhance AI interpretability and foster trust. This initiative emphasized human-machine collaboration and positioned AI as an augmentative tool rather than a replacement. This study highlights the process of designing a human-centered AI for dispensing verification, emphasizing its interpretability, confidence visualization, and collaborative human-machine teaming styles.

A Research Framework Focused on AI and Humans instead of AI versus Humans

Despite lacking a shared understanding and a generally accepted definition, Artificial intelligence (AI) is promoted and credited with miraculous abilities to solve all problems. To gain a more nuanced and deeper understanding of the design trade-offs associated with AI, this paper proposes a research framework that contrasts two competing frameworks: (1) AI versus Humans (characterized by strong AI and Artificial General Intelligence) focused on replacing human beings and (2) AI and Humans (characterized by intelligence augmentation and human-centered AI) focused on empowering human beings as individuals and communities. The arguments in the paper are supported by research activities that explored conceptual frameworks and inspiring prototypes. These developments have resulted in gaining a deeper understanding of how AI-type systems can contribute to quality of life aspects with a specific focus on rethinking and reinventing learning, education, working, and collaboration in the digital age.

Cultivation of human centered artificial intelligence: culturally adaptive thinking in education (CATE) for AI.

Artificial Intelligence (AI) has become ubiquitous in human society, and yet vast segments of the global population have no, little, or counterproductive information about AI. It is necessary to teach AI topics on a mass scale. While there is a rush to implement academic initiatives, scant attention has been paid to the unique challenges of teaching AI curricula to a global and culturally diverse audience with varying expectations of privacy, technological autonomy, risk preference, and knowledge sharing. Our study fills this void by focusing on AI elements in a new framework titled Culturally Adaptive Thinking in Education for AI (CATE-AI) to enable teaching AI concepts to culturally diverse learners. Failure to contextualize and sensitize AI education to culture and other categorical human-thought clusters, can lead to several undesirable effects including confusion, AI-phobia, cultural biases to AI, increased resistance toward AI technologies and AI education. We discuss and integrate human behavior theories, AI applications research, educational frameworks, and human centered AI principles to articulate CATE-AI. In the first part of this paper, we present the development a significantly enhanced version of CATE. In the second part, we explore textual data from AI related news articles to generate insights that lay the foundation for CATE-AI, and support our findings. The CATE-AI framework can help learners study artificial intelligence topics more effectively by serving as a basis for adapting and contextualizing AI to their sociocultural needs.

Looking at human-centered artificial intelligence as a problem and prospect for sociology: An analytic review

Significant advances have been achieved within the past decade in the progress of theoretical and empirical studies of Artificial Intelligence. This article is an attempt, through a review of existing literature on Human-Centered Artificial Intelligence, to raise new questions and provide additional scientific data that will stimulate the potential and foster the forces of sociology and Artificial Intelligence studies to draw closer together. The point of departure for the article is the appearance of Human-Centered Artificial Intelligence in scholarly assembly. The authors then explore routines of the term Human-Centered Artificial Intelligence and the dilemmas of Human-Centered Artificial Intelligence. In what follows, they review how Human-Centered Artificial Intelligence appears in sociological and social sciences production. The authors turn to the closing remarks and finalize formulating three rules of what not to do when studying Human-Centered Artificial Intelligence from a sociological perspective.

Humans and cyber-physical systems as teammates? Characteristics and applicability of the human-machine-teaming concept in intelligent manufacturing.

The paper explores and comments on the theoretical concept of human-machine-teaming in intelligent manufacturing. Industrial production is an important area of work applications and should be developed toward a more anthropocentric Industry 4.0/5.0. Teaming is used a design metaphor for human-centered integration of workers and complex cyber-physical-production systems using artificial intelligence. Concrete algorithmic solutions for technical processes should be based on theoretical concepts. A combination of literature scoping review and commentary was used to identify key characteristics for teaming applicable to the work environment addressed. From the body of literature, five criteria were selected and commented on. Two characteristics seemed particularly promising to guide the development of human-centered artificial intelligence and create tangible benefits in the mid-term: complementarity and shared knowledge/goals. These criteria are outlined with two industrial examples: human-robot-collaboration in assembly and intelligent decision support in thermal spraying. The main objective of the paper is to contribute to the discourse on human-centered artificial intelligence by exploring the theoretical concept of human-machine-teaming from a human-oriented perspective. Future research should focus on the empirical implementation and evaluation of teaming characteristics from different transdisciplinary viewpoints.

What is critical for human-centered AI at work? - Toward an interdisciplinary theory.

Human-centered artificial intelligence (HCAI) has gained momentum in the scientific discourse but still lacks clarity. In particular, disciplinary differences regarding the scope of HCAI have become apparent and were criticized, calling for a systematic mapping of conceptualizations-especially with regard to the work context. This article compares how human factors and ergonomics (HFE), psychology, human-computer interaction (HCI), information science, and adult education view HCAI and discusses their normative, theoretical, and methodological approaches toward HCAI, as well as the implications for research and practice. It will be argued that an interdisciplinary approach is critical for developing, transferring, and implementing HCAI at work. Additionally, it will be shown that the presented disciplines are well-suited for conceptualizing HCAI and bringing it into practice since they are united in one aspect: they all place the human being in the center of their theory and research. Many critical aspects for successful HCAI, as well as minimum fields of action, were further identified, such as human capability and controllability (HFE perspective), autonomy and trust (psychology and HCI perspective), learning and teaching designs across target groups (adult education perspective), as much as information behavior and information literacy (information science perspective). As such, the article lays the ground for a theory of human-centered interdisciplinary AI, i.e., the Synergistic Human-AI Symbiosis Theory (SHAST), whose conceptual framework and founding pillars will be introduced.

Understanding Lived Experience: Bridging Artificial Intelligence and Natural Language Processing with Humanities and Social Sciences

True human-centered Artificial Intelligence (AI) is impossible without addressing the inherent and diverse aspects of humanness. Deep learning models have achieved remarkable success in some tasks in vision and language processing, and few can deny it. However, as it moves forward, the field cannot continue to pretend it can do it all by itself, especially when we advertise it as ‘human-centered AI’. It has come the time to open up the stage for methodological pluralism in the interest of critical and democratic science, and for the benefit of society. In this paper, I want to draw particular attention to the aspect of lived (subjective) experience, one research area highly misunderstood and hugely neglected in AI, and especially in Natural Language Processing (NLP). Our intensions, selfhood, autonomy, emotions, feelings, sensory knowledge, cultural history are integral components of our intelligence. Thus, the future AI and NLP models will need to more closely align with the embodied component of human intelligence. As we push the limit of creativity and innovation in AI, we need to develop a new way of looking at human experience, with a better scientific understanding of intelligence and its own practices, at the intersection of many disciplinary fields.

Public Health Calls for/with AI: An Ethnographic Perspective

Artificial Intelligence (AI) based technologies are increasingly being integrated into public sector programs to help with decision-support and effective distribution of constrained resources. The field of Computer Supported Cooperative Work (CSCW) has begun to examine how the resultant sociotechnical systems may be designed appropriately when targeting underserved populations. We present an ethnographic study of a large-scale real-world integration of an AI system for resource allocation in a call-based maternal and child health program in India. Our findings uncover complexities around determining who benefits from the intervention, how the human-AI collaboration is managed, when intervention must take place in alignment with various priorities, and why the AI is sought, for what purpose. Our paper offers takeaways for human-centered AI integration in public health, drawing attention to the work done by the AI as actor, the work of configuring the human-AI partnership with multiple diverse stakeholders, and the work of aligning program goals for design and implementation through continual dialogue across stakeholders.