The inspiration comes from a discipline far removed from computer science. Engineers stress-test bridges by loading them beyond expected limits; banks stress-test portfolios against market crashes. The researchers translated that philosophy into machine learning by asking a deceptively simple question: not whether a model is fair under the data it happens to receive, but how quickly it stops being fair as the training environment deteriorates. Static fairness audits, the kind most organizations run before deployment, capture a single snapshot. SLF–FST instead produces a movie, tracing the joint trajectory of predictive performance and group-based fairness metrics as bias intensity climbs from zero to twenty percent.

Technically, the framework works by selectively flipping class labels in a way that is conditioned on group membership. The method targets two specific subsets of the training data: instances from the protected group with positive outcomes, and instances from the privileged group with negative outcomes. Flipping the first set converts favorable outcomes into unfavorable ones, while flipping the second does the opposite, systematically inflating the apparent success rate of the privileged group while suppressing that of the protected group. A parameter called the pollution rate controls how many eligible labels are corrupted, ranging from five to twenty percent in the experiments. Crucially, only labels change; the feature distributions remain untouched, so any degradation in model behavior can be attributed directly to the injected bias rather than to artifacts of data manipulation.

The researchers also developed three distinct injection strategies that differ in how instances are selected for corruption. The LOW strategy targets instances where an auxiliary Random Forest estimator is most uncertain, meaning cases sitting near the decision boundary. The HIGH strategy flips labels the estimator is most confident about, directly contradicting the strongest feature-label relationships in the data. The RANDOM strategy selects instances uniformly at random from the eligible subsets, ignoring confidence entirely. Each strategy was designed to test a different hypothesis about how structured bias propagates through a learning algorithm, and the results largely confirmed the team’s expectations.

The experimental scale was substantial. The team trained four classifier families, Decision Tree, Logistic Regression, Random Forest, and a feedforward Neural Network, on three widely used benchmark datasets: Adult, which predicts whether income exceeds fifty thousand dollars; Bank Marketing, which predicts subscription to a financial product; and COMPAS, the notorious criminal recidivism dataset whose racially disparate risk assessments helped ignite the algorithmic fairness debate. Each dataset carries a designated sensitive attribute: gender for Adult, marital status for Bank Marketing, and race for COMPAS. The entire pipeline, including stratified fivefold cross-validation and Bayesian hyperparameter optimization with Optuna, was repeated eight times, producing a total of 6,240 trained models. Notably, no fairness metric was used during hyperparameter tuning, mirroring common real-world development practice and allowing the impact of biased data to be observed in isolation.

The headline finding concerns which algorithms break first. Random Forests, benefiting from the averaging power of an ensemble, proved the most robust to injected bias, degrading the most gracefully as pollution rates rose. Logistic Regression exhibited the highest sensitivity, with the most pronounced decline in both predictive accuracy and fairness on the COMPAS dataset. Neural Networks and Decision Trees landed in the intermediate zone. The explanation for Logistic Regression’s fragility lies in its linear decision boundary: because the model fits a single separating hyperplane, accumulated perturbations to high-confidence instances can force an abrupt, large-scale shift in that boundary, whereas ensembles of trees absorb individual distortions through majority voting.

The three injection strategies produced a striking and somewhat counterintuitive pattern. The HIGH strategy, which contradicts the strongest feature-label associations, caused the most severe damage to predictive performance but comparatively smaller increases in measured unfairness. The LOW strategy did the opposite: it amplified group disparities substantially while leaving overall accuracy largely intact, because the flipped instances occupied ambiguous regions of the feature space where the model was already uncertain. This makes LOW the standout choice for fairness stress testing in practice, since it isolates fairness degradation from accuracy loss and produces deterministic, reproducible results. RANDOM fell between the two, adding variability without the targeting precision of the confidence-guided approaches.

The study also documented subtle, dataset-specific behaviors that a static audit would never catch. On the Bank Marketing dataset, Statistical Parity initially decreased when bias was first injected, because the protected group started with a higher proportion of positive outcomes, and early flips temporarily balanced the two groups before the disparity reversed and grew. On Adult, Logistic Regression under the HIGH strategy showed an S-shaped performance curve, dipping sharply at ten percent pollution before partially recovering as flipped labels became more evenly distributed across classes. In COMPAS, aggressive flipping occasionally disrupted spurious correlations that the linear model would otherwise exploit, counterintuitively improving some fairness metrics. These anomalies matter because they demonstrate that fairness and accuracy interact in non-linear, model-dependent ways that single-point evaluations fundamentally cannot capture.

The practical implications extend well beyond the laboratory. The authors position SLF–FST as a pre-deployment auditing mechanism: by identifying the pollution rate at which a given model’s fairness metrics cross an unacceptable threshold, practitioners can quantify how much label bias their system can tolerate before it fails. This failure threshold is exactly the kind of information that regulators and auditors increasingly demand, and the framework’s model-agnostic design means it can be applied to any classifier without modification. The code and datasets are publicly available on GitHub, and the paper is open access, lowering the barrier for organizations to adopt the protocol.

The researchers are candid about limitations. The confidence-guided strategies depend on an auxiliary estimator, and the results characterize a Random-Forest-guided instantiation specifically; different ranking models could select different samples and shift the degradation trajectories. The study also confines itself to binary classification and a single sensitive attribute per dataset, leaving multi-class settings and intersectional identities for future work. Still, the core message stands with unusual clarity: robustness to biased data is a property that varies dramatically across algorithms, and it cannot be inferred from how a model performs on clean data. As machine learning systems take on higher-stakes decisions, the question is no longer only whether a model is fair today, but whether it can survive the biased world it will actually be trained in. Stress testing, this study suggests, is how we find out before it matters.

Subject of Research: A fairness stress-testing framework that injects progressive group-conditional label bias to measure how classification algorithms degrade in accuracy and fairness.

Article Title: SLF–FST: a framework for stress-testing fairness under progressive label bias

Article References: Pagliusi, R., Alvim, L., Ferreira, R. S., Canalli, Y., Braida, F., & Zimbrão, G. (2026). SLF–FST: a framework for stress-testing fairness under progressive label bias. International Journal of Data Science and Analytics, 22(1), Article 319. https://doi.org/10.1007/s41060-026-01294-4

Image Credits: AI Generated

DOI: 10.1007/s41060-026-01294-4

Keywords: algorithmic fairness, machine learning, label bias, stress testing, COMPAS, Random Forest, Logistic Regression, bias injection, fairness metrics, classifier robustness, pre-deployment auditing, responsible AI

Their system, called React-to-Me, is a conversational assistant built on top of the Reactome Pathway Knowledgebase, one of the most rigorously maintained curated resources in molecular biology. Reactome, stewarded by a team of expert biologists, documents human biological pathways, molecular interactions, and disease mechanisms with a level of care that ordinary web search cannot match. Yet the knowledgebase has a usability problem: its data model is intricate, and its search interface rewards people who already know the vocabulary and structure of the underlying database. For a graduate student, a clinician, or a researcher crossing into a new field, that learning curve can be discouraging. React-to-Me was designed to let such users ask questions in plain natural language while still receiving answers that are traceable, verifiable, and anchored in curated content.

The engineering heart of the system is a hybrid retrieval-augmented generation pipeline, a technique that constrains a language model to answer only from documents fetched at query time rather than from its internal memory. In React-to-Me, an incoming question triggers two complementary searches over the Reactome corpus. The first is a semantic vector search, which uses neural embeddings to match the meaning of the question against the meaning of database entries, capturing paraphrases and conceptual overlap that keyword matching would miss. The second is a lexical keyword search, of the BM25 variety familiar from classic information retrieval, which excels at pinning down precise identifiers, gene names, and chemical terms. The two ranked lists are merged with reciprocal rank fusion, a method that rewards documents appearing near the top of both lists, and the fused evidence is handed to a language model that is instructed to generate an answer grounded strictly in that retrieved material.

Grounding alone is not enough; the system also insists on provenance. Every response is directly linked to the corresponding Reactome entries from which it was assembled, so users can click through and inspect the curated pathways behind the answer. And when the retrieved evidence is thin or absent, React-to-Me is designed to defer rather than improvise: it declines to speculate and instead points users toward trusted external biomedical sources. This refusal behavior, the authors argue, is a feature rather than a failure, because an honest admission of limited coverage is far more valuable to a scientist than a fabricated pathway. The team tested how well that refusal logic worked in the wild by manually classifying logged instances where the system declined to answer, finding that out-of-scope biology questions were the largest category, accounting for 31.9 percent of refusals.

Before any human saw the system, the researchers put the retrieval engine itself through computational benchmarking using the Ragas evaluation framework, scoring configurations on context utilization, relevance, and faithfulness across several query types: simple factual lookup, mechanistic reasoning, multi-context synthesis, and conditional logic. The results showed a clear pattern. Semantic retrieval on its own preserved relevance but faltered on faithfulness, letting unsupported statements slip through, while pure keyword retrieval weakened on complex reasoning tasks that required synthesizing multiple sources. The combined hybrid approach, ranked by reciprocal rank fusion, delivered the highest and most stable scores across every metric and every query type, confirming that the two retrieval strategies compensate for each other’s blind spots.

Real-world deployment brought a second wave of evidence. Analyzing user queries logged during October 2025 and March 2026, the team found that factual lookup and mechanistic or relational biology questions dominated, together accounting for 56.1 percent of classified logged-in queries. Notably, the mix shifted over time: factual lookups rose from 24.7 percent to 33.7 percent of classified queries, while mechanistic and relational questions declined from 32.7 percent to 21.2 percent, a pattern suggesting that as users became comfortable with the tool, they leaned on it increasingly as a rapid reference rather than a reasoning partner. The study received ethics approval from the University of Toronto Research Ethics Board under protocol number 47192, and all survey participants provided informed consent.

The most demanding test was a blinded head-to-head comparison against a general-purpose model, GPT-4o-mini, judged by ten external molecular biology experts who did not know which system produced which answer. The verdict favored grounding decisively. Grounded React-to-Me responses were more likely to receive higher quality ratings than their ungrounded counterparts, with an overall odds ratio of 2.01. The gains were strongest where scientific rigor matters most: factual accuracy improved with an odds ratio of 2.97, biological specificity with an odds ratio of 2.88, and mechanistic depth with an odds ratio of 1.83. Individual evaluator preferences, recorded in the supplementary data, showed consistent majority favoring of the grounded system across question-level comparisons, an exploratory mixed-effects ordinal regression accounting for variation among evaluators and questions.

User sentiment reinforced the expert verdict. In a survey of 25 consenting users drawn from the publicly deployed system, 92 percent expressed strong satisfaction with ease of use, 88 percent with citation reliability, and 85 percent with factual accuracy. Strikingly, perceived accuracy showed the strongest association with overall confidence in the system, with a correlation of r = 0.81, meaning that users who trusted the answers did so primarily because the facts checked out, not because the interface was pretty. Respondents spanned organization types and research fields, with human biology research most represented, followed by computational biology, cell biology, and genetics, and satisfaction held up across education levels and levels of prior Reactome experience.

Beyond the headline numbers, the study offers a practical template for institutions wrestling with how to deploy AI responsibly. The team published monthly operating costs, the full survey instrument, benchmark questions, and analysis code as supplementary materials, an unusual degree of transparency for a production AI deployment. The willingness to defer to trusted sources when coverage runs out, the insistence on clickable citations for every claim, and the hybrid retrieval architecture that balances conceptual breadth with lexical precision together sketch what credible scientific AI might look like. As funders and publishers grapple with hallucination risks in research tooling, React-to-Me demonstrates that domain-specific grounding is not a theoretical nicety but a measurable design choice, one that nearly triples the odds that an expert will judge an answer factually accurate. The broader lesson may extend well beyond pathways: for specialized knowledge, the future of conversational AI belongs to systems that know the boundaries of what they can honestly say.

Subject of Research: A grounded conversational AI interface for querying the Reactome Pathway Knowledgebase

Article Title: React-to-Me: real-world experience of a grounded conversational interface to the Reactome Pathway Knowledgebase

Article References: Mohammadi, H., Almodaresi, F., Hogue, G. F. J., Wright, A., Orlic-Milacic, M., Li, N. T., Mawani, A., & Stein, L. (2026). React-to-Me: real-world experience of a grounded conversational interface to the Reactome Pathway Knowledgebase. BMC Biology. https://doi.org/10.1186/s12915-026-02740-2

Image Credits: AI Generated

DOI: 10.1186/s12915-026-02740-2

Keywords: Reactome Pathway Knowledgebase, conversational AI, retrieval-augmented generation, large language models, knowledge grounding, biological pathways, biomedical question answering, hybrid retrieval, language model evaluation, computational biology, expert evaluation, responsible AI

At a technical level, generative AI differs from the analytical machine learning that retailers have used for more than a decade. Traditional predictive models classify or forecast: they estimate the probability that a customer will click, churn or respond to a discount. Generative models, by contrast, learn the underlying distribution of their training data and can produce new content that resembles it, including text, images, product descriptions, conversational responses and synthetic demand scenarios. This capability is what enables many of the applications the book examines. AI-powered personalization, for example, can move beyond static segment-based recommendations toward dynamically generated content tailored to an individual shopper’s context, browsing history and stated preferences in real time. Recommendation systems, long a staple of e-commerce platforms, can be augmented with language models that explain why an item is suggested, answer follow-up questions and negotiate the trade-offs between relevance, diversity and margin.

One of the most prominent themes in the volume is conversational commerce, the use of natural language interfaces to mediate the shopping journey. Virtual shopping assistants built on large language models can interpret free-form customer queries, clarify ambiguous requests, compare products across attributes and guide users through complex purchases such as electronics, furniture or fashion ensembles. The book’s treatment of this topic reflects a broader consensus among practitioners that the search box and the filter menu are being supplemented, and in some cases replaced, by dialogue. Because generative models can handle unstructured inputs, they reduce the friction that historically forced customers to translate their needs into keyword queries. The contributors also address the engineering realities behind such systems, including retrieval-augmented generation, in which a language model is grounded in a retailer’s live product catalog and policy documents to reduce factual errors and hallucinated specifications.

Beyond the customer-facing surface, the book devotes substantial attention to operational applications where generative and predictive techniques converge. Demand forecasting has traditionally relied on statistical time-series methods that extrapolate from historical sales, seasonality and promotional calendars. Modern systems increasingly combine these with machine learning features such as weather, local events and social signals, and generative models can now simulate plausible future demand scenarios to stress-test inventory plans. Intelligent supply chains, another chapter theme, use similar reasoning: AI can generate candidate sourcing strategies, optimize replenishment schedules and draft contingency plans for disruption. Retail analytics more broadly benefits from generative capabilities that summarize performance data in natural language, allowing managers without data-science training to interrogate dashboards conversationally and receive narrative explanations of anomalies.

Merchandising and pricing represent a particularly fertile area for generative methods. The book examines how AI can generate product copy, imagery and campaign assets at scale, compressing creative production cycles that once took weeks into hours. In pricing, generative agents can be paired with optimization engines to explore how customers might respond to alternative price points, promotions and bundle configurations before they are deployed, effectively running low-cost simulations of market behavior. The editors emphasize that these capabilities come with governance requirements: automated pricing and content generation must remain within legal and ethical boundaries, avoid discrimination, and preserve brand voice, which is why the volume pairs its technical chapters with discussions of ethical and responsible AI adoption.

The structure of the book is designed to serve both academic and practitioner audiences. Organized into fourteen chapters, it moves from foundational concepts and theoretical perspectives through practical applications, industry case studies and future trends in AI-enabled retailing. Contributions come from scholars and practitioners, a combination the editors describe as providing both academic rigor and real-world insight. Contemporary case studies, practical frameworks, evidence-based research findings and strategic recommendations appear throughout, and every chapter carries references, giving readers entry points into the underlying literature. Key features highlighted by the publisher include interdisciplinary perspectives on AI applications in retail and structured content that integrates theory and practice while pointing toward future directions for research and innovation in smart commerce.

Customer experience management receives sustained treatment as the connective tissue linking these technologies. The editors argue that the ultimate measure of any generative deployment is not model performance in isolation but its effect on the end-to-end journey: whether personalization feels helpful rather than intrusive, whether conversational assistants resolve issues without escalating to human agents, and whether omnichannel experiences remain consistent as customers move between web, mobile, physical stores and social commerce. Sustainability is also framed as part of this value equation rather than an afterthought. AI-driven forecasting and inventory optimization can reduce overproduction and waste, while generative design tools can support more efficient packaging and logistics planning, aligning commercial and environmental objectives.

The editorial team brings together five academics with complementary specializations. Nupur Arora and Aanchal Aggarwal are based at the School of Business Studies of Vivekananda Institute of Professional Studies–TC in New Delhi, India. Parul Manchanda is affiliated with the Department of Management Studies at Netaji Subhas University of Technology, also in New Delhi. Rohit Bansal works in the Department of Management at Rockford College in Sydney, Australia, and Ramakrishna Yanamandra is at the School of Business of Horizon University College in Ajman, United Arab Emirates. This geographic and disciplinary spread is reflected in the book’s scope, which spans marketing, retail management, business analytics, information systems and artificial intelligence research traditions.

The publisher identifies a primary readership of researchers, academicians, doctoral scholars and postgraduate students in those same fields, and a secondary audience of retail managers, business leaders, consultants, entrepreneurs, technology professionals, policymakers and industry practitioners seeking to understand and implement AI-driven retail innovations. That dual orientation matters at a moment when the gap between academic research and commercial practice in AI can be wide. Frameworks that survive peer review do not always translate into deployable systems, and vendor claims rarely come with rigorous evaluation. A reference work that aggregates case studies and evidence-based findings from both sides offers a middle path: decision-makers gain a vocabulary for assessing what is technically feasible, while researchers gain visibility into the operational constraints that shape real deployments.

The broader significance of the book lies in its timing. Generative AI has moved from research laboratories to production systems in retail at remarkable speed, and the industry is still developing the norms, evaluation methods and regulatory awareness needed to deploy it responsibly. Questions the volume engages with, including hallucination in customer-facing assistants, the provenance of AI-generated content, the fairness of algorithmic pricing and the labor implications of automation, are now live policy debates in multiple jurisdictions. By consolidating foundational concepts, applied evidence and future trends into a single reference, Generative AI for Retail Innovation positions itself as a map of a field that is being written in real time, useful both as a teaching resource and as a strategic guide for organizations navigating the transition to AI-enabled commerce and sustainable business growth.

Subject of Research: Applications of generative artificial intelligence in retail operations and strategy

Article Title: Generative AI for Retail Innovation

Article References: Generative AI for Retail Innovation. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: generative AI, retail, personalization, conversational commerce, recommendation systems, demand forecasting, supply chain, omnichannel retailing, retail analytics, responsible AI, sustainability, book release

The study arrives at a moment when AI-based literature has gained unprecedented traction, particularly since 2022, and when automated decision-making has spread across business functions from human resource management to customer service platforms. Vulnerable and minority employees frequently face systemic barriers that include discrimination, underrepresentation in leadership positions, limited access to career advancement, and insufficient organisational support. AI-based recruitment, performance evaluation, and career development systems promise to reduce these inequities by emphasising standardised, data-driven decisions that minimise the influence of unconscious human prejudice. The technology can also support inclusive job matching, skill development, and accessibility initiatives, empowering workers who have traditionally experienced disadvantage. Yet poorly designed or unmonitored systems risk creating feedback loops that systematically disadvantage the very groups they could help, as high-profile cases of recruitment algorithms reproducing gender and racial bias from biased historical data have demonstrated.

To map this contested terrain, the review employed an unusually rigorous multi-phase methodology. The author conducted a keyword-based literature search in the Scopus database on August 10, 2025, using a Boolean query that combined terms for artificial intelligence, machine learning, and algorithmic decision-making with terms covering vulnerable and minority employee populations, human resource management, and ethics. The initial search identified 554 articles; after excluding 308 records that were not the required publication types and 9 articles not in English, 237 articles entered bibliometric mapping and principal component analysis. A final manual thematic synthesis drew on 30 articles published in A*, A, and B-ranked journals according to the Australian Business Deans Council index. The full PRISMA-guided screening process was documented to ensure transparency, and all 237 articles were included in the statistical analyses to capture the interdisciplinary breadth of the field, spanning information systems, AI ethics, disability studies, and organisational behaviour.

The quantitative core of the review combined VOSviewer bibliometric mapping with principal component analysis. Keyword co-occurrence analysis identified 93 recurring keywords, and a Kaiser-Meyer-Olkin test of sampling adequacy, with a threshold of 0.50, filtered these down to 16 keywords for the final PCA. Applying Kaiser’s criterion of eigenvalues greater than 1, the analysis extracted six principal components, which the author labelled as social equity and representation, ethical governance and AI technology, inclusion and organisational adaptation, employment and knowledge transformation, structural challenges in workforce diversity, and goals and workplace realities. Scree plots illustrated the effect of the dimensionality reduction. These statistically derived components were then consolidated, through qualitative interpretation, into five overarching themes that structure the review’s synthesis: AI adoption and workforce transformation; bias, equity, and fairness in AI systems; employee well-being, inclusion, and empowerment; ethical, legal, and governance considerations; and methodological approaches and tools.

Publication trends reveal how young the field is. Minimal contributions appeared between 2006 and 2019, but a sharp upward trajectory began in 2020, with publications rising to six that year and eight by 2023. Knowledge Management Research and Practice leads the top ten journals by total citations with 305, followed by Informing Science with 183 and the Journal of Information, Communication and Ethics in Society with 167. The theoretical landscape underpinning the literature is rich but fragmented, and the review groups it into four domains: ethics, justice, and fairness; organisational and human resource management; technology and digital transformation; and disability, diversity, and inclusion frameworks. The first domain includes procedural justice, algorithmic fairness, and responsible AI innovation models. The second draws on the Resource-Based View, the Dynamic Capability Framework, and strategic human resource management perspectives. The third relies on technology adoption and algorithmic management theories, while the fourth deploys frameworks such as Disability Justice, Feminist Design Thinking, neurodiversity models, and identity-consciousness versus identity-blindness approaches that position marginalised employees not merely as subjects of algorithmic governance but as knowledge holders and co-designers of inclusive AI systems.

Among the review’s most striking empirical findings is evidence that AI-generated communications can provoke stronger negative reactions toward employees with disabilities or women than traditional human-based bias, exceeding it in some contexts. Research on construction and engineering leadership documents a likeability versus competency dilemma, in which women with comparable qualifications and experience are perceived as less likeable than male peers. In digital skills and STEM training programmes, recruiters using AI-based candidate screening were found to favour male candidates during initial outreach, especially under high workloads, demonstrating how such systems can reinforce gender-based disparities even before applications are submitted. Studies of AI-assisted disability assessments show that biases may emerge from design choices, data selection, or operational deployment, underscoring the importance of participatory, disability-led design practices. Meanwhile, profiling models in public employment services can inadvertently misclassify and discriminate against minority or foreign-origin jobseekers, exposing a sharp trade-off between accuracy and equity.

The review also documents AI’s genuine potential for empowerment. AI-driven innovation can enhance transparency, strengthen internal controls, and shape workplace culture and performance evaluation systems, while applications in policy evaluation, exemplified by China’s Low-Carbon City Pilot program, show how algorithmic tools can improve job quality, entrepreneurship opportunities, and urban labour inclusivity. Healthcare studies demonstrate that AI systems designed with clear reasoning, adaptive triage, and data transparency can reduce cognitive burdens while promoting equitable outcomes for diverse employees. Research on workforce diversity, equity, and inclusion in healthcare further indicates that improvements across demographic and experiential dimensions correlate with better patient safety outcomes, particularly in regions with diverse patient populations. Studies of corporate diversity statements show that companies emphasising identity-conscious topics receive more favourable employee evaluations of DEI, and a tri-balance framework for AI in personnel selection illustrates how efficiency, fairness, and stakeholder voice might be reconciled.

From these threads the review advances an integrative framework built on three interacting dimensions: the technological mechanisms of AI systems, organisational and governance mediators, and employee outcomes. The framework’s central claim is that empowerment outcomes are not determined solely by the technologies themselves but by the interaction between algorithmic design, organisational implementation practices, and governance structures. When transparency, fairness auditing, and inclusive organisational policies are implemented appropriately, AI systems may support more equitable and empowering workplace environments. When they are not, the same systems can entrench exclusion. This reframing positions AI not as a neutral technical artefact but as a sociotechnical governance issue shaped by organisational values and institutional structures, an account that aligns closely with emerging information systems scholarship on responsible AI and algorithmic governance. The author also notes, however, that the field’s empirical contributions remain fragmented and uneven, with equity research often disconnected from employee experiences and methodological innovations rarely integrated into discussions of ethics or well-being.

The review acknowledges its own limitations, including reliance on a single database, the imperfect reach of keyword-based searches, the possible influence of journal quality filters on corpus composition, and the inherent interpretive judgments of qualitative synthesis. Yet its research agenda is ambitious. The author calls for future work to connect social role and congruity theories with strategic human capital and corporate governance frameworks, to combine feminist design thinking with identity-consciousness debates, and to integrate neurodiversity and disability-led design perspectives with digital transformation research. On the empirical side, the review urges multi-country, longitudinal, and high-dimensional fixed-effects models to capture institutional, cultural, and regulatory heterogeneity, noting that labour laws, political climate, and social norms may shape how AI and DEI initiatives are implemented and received. With most current studies relying on cross-sectional or single-country designs, the consequences, rather than merely the determinants, of AI-augmented diversity management remain largely unexplored. What is already clear, the review concludes, is a fundamental tension at the heart of workplace AI: whether the technology becomes an instrument of algorithmic exclusion or a genuine engine of empowerment will depend on choices, about design, oversight, and governance, that organisations are making right now.

Subject of Research: A systematic review of how artificial intelligence affects the empowerment, equity, and inclusion of vulnerable and minority employees in the workplace

Article Title: Artificial Intelligence and the Empowerment of Vulnerable and Minority Employees: A Systematic Review

Article References: Pokharel, P. R. (2026). Artificial Intelligence and the Empowerment of Vulnerable and Minority Employees: A Systematic Review. Information Systems Frontiers. https://doi.org/10.1007/s10796-026-10823-2

Image Credits: AI Generated

DOI: 10.1007/s10796-026-10823-2

Keywords: artificial intelligence, vulnerable employees, minority employees, algorithmic bias, diversity equity and inclusion, systematic review, human resource management, algorithmic governance, employee empowerment, workplace transformation, responsible AI, information systems

The scale of the underlying problem is enormous. Nearly sixty million people worldwide live with dementia, a syndrome most commonly caused by Alzheimer’s disease and marked by progressive deterioration of memory, speech and judgment, and that figure is expected to more than double by 2050. Because irreversible brain damage typically occurs long before a clinical diagnosis is made, the most effective defence against the disease is early detection paired with risk-reducing lifestyle interventions. Yet conventional diagnostic approaches, which combine psychometric testing, imaging and lifestyle data, are time- and resource-intensive, often requiring multiple specialist referrals and producing delays that were further exacerbated by the COVID-19 pandemic. Simplified toolkits such as the CAIDE and BDSI risk scores were designed to help, but they operate over fixed, pre-specified horizons, target narrow age groups, and in some analyses add little discriminative value beyond knowing a patient’s age alone.

Previous machine learning models trained on large clinical datasets promised improvements but were typically locked into a single fixed prediction window, such as twenty-nine months or one year. The research team behind the new study hypothesised that this rigidity is itself a core limitation: defining one prediction horizon discards valuable data and can introduce sampling bias when right-censored data, meaning patients whose follow-up ended before the horizon, is handled poorly. Their solution was elegant. They introduced a feature called HORIZON, which encodes the number of months between a patient’s baseline visit and each subsequent visit, and set it to zero for immediate diagnosis. Rather than building separate models for each time frame, the algorithm treats the prediction window as just another input, learning how the relationship between risk factors and dementia changes, or does not change, as the horizon stretches out. The approach is conceptually related to a statistical technique known as survival stacking, which recasts time-to-event analysis as a classification problem, but here the researchers exploited the naturally rich multi-visit structure of their dataset instead of generating synthetic samples.

The data came from the Uniform Data Set of the National Alzheimer’s Coordinating Center, a collaboration spanning more than 42 American research centres, covering 47,400 patients between June 2005 and February 2023. After rigorous preprocessing, including the removal of variables with more than seventy percent missing values and the elimination of highly correlated features, the team arrived at 134,117 visit records from 47,017 unique patients. Critically, variable selection was clinically informed from the outset: a consultant psychiatrist working in an NHS memory assessment service reviewed candidate variables against two criteria, routine availability in secondary care and minimal reliance on subjective expert judgment, so that the final inputs could plausibly be collected by junior doctors or medical assistants. Recursive feature elimination, embedded within a nested cross-validation pipeline, then narrowed the field to the twenty most informative variables.

Six machine learning algorithms competed in the study, and the K-Nearest Neighbour classifier emerged as the strongest. Its logic is intuitively clinical: to classify a new patient, the model compares their profile against the hundred most similar patients with known dementia status and weighs each equally, effectively mimicking a clinician reasoning from a wealth of prior comparable cases. For same-visit diagnosis, the model achieved a sensitivity of 0.951 and a specificity of 0.810, with an area under the receiver operating characteristic curve of 0.959. Across future horizons from 24 to 120 months, sensitivity remained high, ranging between 0.808 at two years and 0.833 at ten years, while specificity declined from 0.856 to 0.706 as the horizon lengthened, an expected trade-off in risk prediction. Class imbalance was addressed with SMOTE-NC oversampling applied within each horizon group, and all preprocessing was confined to training folds to prevent information leakage.

The flexible-horizon model was also benchmarked against a Cox proportional hazards survival model, the conventional statistical approach for time-to-event outcomes. The comparison revealed complementary strengths: the KNN model achieved a higher geometric mean of sensitivity and specificity at four of the five evaluated horizons and substantially greater specificity from 48 months onwards, whereas the Cox model delivered greater sensitivity at longer horizons. Importantly, the machine learning approach offers practical advantages beyond raw performance, supporting point-of-assessment diagnosis, something a survival model cannot do, and fitting into standard supervised learning pipelines without requiring assumptions such as proportional hazards or non-informative censoring.

Transparency was built into the analysis through Shapley additive explanations, a technique that assigns each feature a contribution value for every individual prediction. Globally, the two most influential features were functional independence and the prediction horizon itself, followed by cognitive measures including the animal naming test, the Mini-Mental State Examination and the geriatric depression scale. The explanations aligned with established clinical understanding: lower cognitive test scores, loss of functional independence, poorer recall of significant dates and difficulty assembling tax records all pushed the model’s risk estimate upward. Local explanations, presented as waterfall plots, showed how individual features shifted the risk score for specific patients, the kind of case-by-case interpretability that clinicians increasingly demand before trusting an algorithm. Notably, age ranked lower than expected, likely because its effect is partially absorbed by the horizon feature.

Fairness received equally serious attention. The team evaluated performance across subgroups defined by sex, race and age, computing equal opportunity, predictive equality and a novel G-mean equality ratio. Balanced discrimination proved broadly similar across most groups, but the trade-offs differed: males showed higher sensitivity but lower specificity than females, Asian and Other race subgroups showed high sensitivity with lower specificity, and the small 18-44 age group showed lower sensitivity but very high specificity. The authors are careful to frame these results as an initial internal fairness assessment rather than proof of equity, noting that some subgroup estimates rested on small sample counts, and they caution that false negatives and false positives carry different clinical consequences, from delayed assessment to unnecessary anxiety and referral burden.

The researchers have deployed the model as an interactive web application built with Streamlit, intended strictly for demonstration, clinician feedback and future validation rather than autonomous diagnosis. The app runs password-protected on a small cloud instance, retains no user data after each session, and has not yet undergone formal cybersecurity assessment, clinical safety review or regulatory evaluation. They estimate that collecting the model’s twenty inputs, many of which are brief: the MMSE takes roughly five to ten minutes, the short-form geriatric depression scale five to seven, and the animal naming task just sixty seconds, could be completed in around thirty minutes within a suitably prepared secondary-care workflow, compared with a conventional memory clinic appointment of an hour and a half or more. Because the feature set includes modifiable risk factors such as blood pressure, body mass index and depression measures, clinicians could in principle use local explanations to simulate how changing specific factors might alter a patient’s projected risk.

The team is candid that the model is not clinically ready. All training and validation data originate from the American NACC cohort, which may not fully represent demographics, referral pathways or assessment procedures in UK secondary care or other health systems. The raw KNN outputs are not fully calibrated as absolute probabilities, particularly at shorter horizons, and should be interpreted as model-estimated risk scores. The model also treats all-cause dementia as a single label, collapsing distinctions between Alzheimer’s, vascular and other subtypes. The authors’ staged pathway forward is methodical: retrospective evaluation on historical data from a real NHS memory clinic, silent prospective testing in which outputs are generated but not acted upon, clinician usability assessment, workflow impact analysis, and finally governance, cybersecurity and regulatory review. If those hurdles are cleared, a flexible-horizon, explainable and fairness-tested algorithm could become a genuine decision-support companion, one that complements rather than replaces clinical judgment, concentrates specialist time on the hardest cases, and helps bring dementia detection closer to the moment when intervention still matters most.

Subject of Research: A flexible-horizon machine learning model for dementia early detection and risk prediction in secondary care using responsible AI principles

Article Title: A Flexible-Horizon Clinical Decision Support Model for Dementia Early Detection and Risk Prediction in Secondary Care: A Responsible AI Approach

Article References: Ez-zizi, A., Seelam, K., Leggett, L., & Malik, B. R. (2026). A Flexible-Horizon Clinical Decision Support Model for Dementia Early Detection and Risk Prediction in Secondary Care: A Responsible AI Approach. Journal of Medical Systems, 50(1), Article 132. https://doi.org/10.1007/s10916-026-02458-2

Image Credits: AI Generated

DOI: 10.1007/s10916-026-02458-2

Keywords: dementia, machine learning, responsible AI, clinical decision support, risk prediction, NACC Uniform Data Set, K-nearest neighbours, SHAP explainability, algorithmic fairness, secondary care, survival analysis, early detection

The starting point of the paper is a familiar tension in modern policing. Public-safety agencies already juggle fragmented, time-sensitive information streams—radio calls, video feeds, sensor alerts, crowd movement, and human reports—yet most of the technology available to them remains retrospective. Footage is reviewed after an incident, logs are examined once a response has been triggered, and pattern analysis is applied only after historical trends have formed. Recent surveys of agentic artificial intelligence show the field moving toward adaptive reasoning over complex goals, and affective-computing research has matured in estimating probabilistic signals related to arousal, stress, vocal intensity, gesture, and group movement. Drones add mobility, altitude, and rapid redeployment, making them attractive sensing platforms. Together, these capabilities make anticipatory public-safety support technically conceivable.

But the authors argue that the shift from detecting incidents after escalation to sensing conditions that may precede escalation cannot be judged by technical performance alone. A raised voice, a compressed crowd, a sudden gesture, or a cluster of bystanders can accompany conflict, yet the same cues may reflect celebration, unrelated stress, disability-related behavior, cultural expressiveness, or simply the ordinary density of urban life. When an aerial system interprets such cues, the error is not confined to a model output: it may redirect police attention, change how officers approach a scene, or make members of the public feel watched, classified, and pre-judged. Because false positives can disproportionately burden communities already subject to heavier surveillance, the paper treats governance as an architectural requirement rather than an afterthought.

At the heart of SkySentience is the Pre-Incident Escalation Index, or PEI, an interpretable advisory score that aggregates four bounded, time-indexed channels: visual agitation indicators such as abrupt gestures and repeated encroachment into personal space; non-lexical acoustic markers such as rising vocal energy and speech overlap, with no attempt to recognize words, intent, or identity; crowd-dynamics measures such as localized compression, slowing or reversing pedestrian flow, and clustering; and a contextual prior built only from documented, non-individualized variables such as venue type, scheduled events, and time-dependent crowding baselines. The index is computed as a weighted linear combination, PEI equals alpha times the visual score plus beta times the acoustic score plus gamma times the crowd score plus delta times the contextual prior. The weights are deliberately left as calibration targets rather than universal constants, because their relative informativeness varies across venues and because the weighting choice is itself a governance decision that should be documented, version-controlled, and reviewed like any other policy parameter.

Because escalation is dynamic, the framework also tracks the short-term change in the index, delta PEI, so that a moderate score rising quickly can attract more attention than a higher but stable score in a benign setting. Every modality must also emit a quality or confidence value reflecting sensor conditions and model reliability, and raw scores should be calibrated on held-out, deployment-relevant data using auditable methods such as isotonic regression or Platt scaling, with expected calibration error and reliability diagrams reported. Missingness is handled conservatively: a degraded channel is never silently imputed, and as a reference rule at least two independent sensing modalities of acceptable quality must be available before any PEI-based advisory is produced; otherwise the system abstains and reports insufficient evidence. The contextual prior does not count as an independent sensor modality, and protected characteristics, inferred identity, and raw historical enforcement frequencies are excluded from the prior because they could reintroduce historical policing patterns into an ostensibly behavior-based score.

To make the design concrete, the paper walks through a hypothetical observation at a busy transit hub in which the visual channel reports 0.60, the acoustic channel 0.48, crowd dynamics 0.70, and the contextual prior 0.20. With documented weights of 0.30, 0.25, 0.30, and 0.15, the PEI works out to 0.54, and against a previous value of 0.43 the trend is plus 0.11. Weighted by quality values from each channel, the aggregate evidence reliability comes to 0.79—a figure the authors stress is an indicator of evidence quality, not a 79 percent probability that escalation will occur. The system would then report the triplet of score, trend, and confidence, and, if validated thresholds were met, issue a plain-language advisory such as a note that the score is moderate and increasing, driven primarily by crowd compression and visual movement cues, recommending continued observation or human review. The authors emphasize that every number in the example is illustrative, not empirical.

The decision logic built on top of the index is intentionally narrow. When the score is low and stable, the drone continues passive observation. When the score is moderate or rising quickly, the system may recommend light-touch, reversible responses such as repositioning for a clearer view or notifying an officer with a short explanation. Even when risk remains elevated and corroborated across modalities, the system prioritizes officer awareness rather than autonomous action, and it is never permitted to stop, identify, pursue, confront, or otherwise act coercively toward members of the public. Language models, if used at all, are restricted to summarizing and explaining policy-checked outputs; they do not control the drone, select enforcement actions, or generate unverified factual claims. The architecture separates perception, advisory reasoning, human authorization, and accountability into distinct layers, so that every action affecting a person’s liberty, safety, or legal exposure requires an explicit human decision.

The governance requirements are equally specific. The authors call for tamper-evident, time-stamped, access-controlled audit logs recording each PEI computation, the contributing modality scores, confidence estimates, recommendations, officer responses, and rationales for acceptance, modification, override, or inaction. They propose an institutional division of labor in which the deploying agency documents deployment criteria and configurations, an independent oversight function outside the chain of command audits logs, override patterns, and subgroup error rates, and a separate public-facing channel handles redress requests from people who believe they were observed or flagged. Proportionality ties intervention intensity to evidentiary strength, confidence, and persistence, while explainability must serve two audiences: officers need concise operational explanations of what drove a score, and the public needs to know what the system observes, what it does not infer, how data are retained or deleted, and how to challenge an assessment. The framework is also jurisdiction-contingent: where law prohibits emotion inference in policing, any channel constituting such inference must be disabled or the deployment must not proceed.

Two illustrative scenarios show how the framework behaves when it is right and when it is wrong. In the first, rising voices, prolonged eye contact, and slowing pedestrian flow at a transit platform jointly lift the index, prompting only a respectful vantage point and a notification to a nearby officer, who retains full discretion to observe, approach, or stand down. In the second, a group celebrating a cultural occasion sings loudly and clusters tightly, pushing the score across a moderate threshold even though the behavior is benign; the system may only reposition or notify, the officer recognizes the celebration and records an override, and that override feeds later recalibration and bias review. The authors argue that the difference between a harmless false positive and a harmful one lies precisely in this governance layer: without bounded autonomy, override logging, and public redress, the same score could produce unnecessary police attention or opaque records.

The paper closes with a candid reckoning. The authors acknowledge the strongest objection—that affect-inferring drone systems should not exist in public policing at all—and they distinguish the justificatory question of whether a community should permit such a system from the architectural question of what must be true of it if deployed. Safeguards, they note, can reduce harm only if a system is deployed; they do not prove deployment is justified, and that prior question belongs to legal, democratic, and community processes. A staged validation plan follows: first testing the index on public benchmark data with subgroup-disaggregated performance, then using simulation and replay studies to examine officer-facing explanations, alert burden, and override patterns, and finally evaluating whether audit records, public explanations, retention policies, and community review mechanisms actually function in practice. A system that passes the technical stage but fails the institutional one, the authors conclude, should not be considered successful—and satisfying every requirement would still not, by itself, establish that any jurisdiction should deploy affect-related drone decision support.

Subject of Research: A conceptual governance-centered framework for accountable, human-supervised UAV decision support in public safety using a Pre-Incident Escalation Index.

Article Title: The SkySentience conceptual framework for accountable drone decision support in public safety

Article References: Majumdar, S., & Awasthi, A. (2026). The SkySentience conceptual framework for accountable drone decision support in public safety. Discover Informatics, 1(1), Article 16. https://doi.org/10.1007/s44564-026-00018-x

Image Credits: AI Generated

DOI: 10.1007/s44564-026-00018-x

Keywords: SkySentience, public-safety informatics, drones, UAV surveillance, agentic AI, Pre-Incident Escalation Index, affect inference, algorithmic accountability, human-AI collaboration, auditability, responsible AI, policing technology

The research team argues that most current scholarship on artificial intelligence in the workplace has been captivated by performance metrics—productivity gains, cost reductions, service quality improvements—while treating the human cost as an afterthought. Their paper shifts the lens. By framing employee psychological wellbeing as a pillar of human sustainability, they contend that an organization cannot be truly sustainable if the mental health, motivation, and social connectedness of its workers deteriorate as algorithms take over more of the daily workload. The paradox emerges because the same adoption decisions that streamline operations can strip away precisely the experiences that make work psychologically nourishing.

To explain how this erosion happens, the authors build their framework on two well-established pillars of organizational psychology: self-determination theory and recovery theory. Self-determination theory holds that human flourishing at work depends on the satisfaction of three basic psychological needs—autonomy, competence, and relatedness. Autonomy is the sense of volition and control over one’s actions; competence is the feeling of effectiveness and mastery; relatedness is the experience of meaningful connection with others. Recovery theory, meanwhile, explains how workers replenish mental and emotional resources during and after demanding periods, and how chronic demands without adequate recovery produce strain, exhaustion, and psychological distress.

Applied to AI adoption, the theoretical machinery generates a sobering prediction. When intelligent systems automate decisions, standardize workflows, and monitor performance, employees may find their discretion shrinking—their sense of autonomy diminished as algorithmic recommendations dictate the pace and content of their work. Competence can suffer when the skills workers spent years honing are rendered obsolete or when success depends on opaque machine outputs they cannot fully understand. Relatedness may weaken as human interactions are replaced by automated transactions and as workers compete with, rather than collaborate alongside, machine agents. The cumulative result, according to the model, is unsatisfied psychological needs, which translate into psychological distress and a decline in what the authors term human sustainability.

Crucially, the framework does not treat this damaging trajectory as inevitable. The researchers identify two factors that shape how harshly AI adoption bears down on workers: employee AI literacy and the prevailing culture of AI within the organization. AI literacy—the knowledge and skills that allow workers to understand, evaluate, and work effectively with intelligent systems—can either amplify or buffer the paradox. Workers with low literacy may feel threatened, confused, and helpless in the face of AI, accelerating the erosion of competence and autonomy. But the relationship is not straightforward, and the authors emphasize that even high literacy alone does not guarantee positive outcomes if the surrounding organizational culture treats AI purely as a tool for surveillance and cost cutting.

This is where human-centered AI practices enter the model as moderators—variables that can soften or even neutralize the negative pathways. The study highlights four in particular. Transparency means that workers understand what the AI does, how it reaches recommendations, and what data it uses, reducing the anxiety born of opacity. Employee involvement means that staff participate in designing, selecting, and configuring AI systems, which preserves a sense of agency and ownership. AI literacy training equips workers with the competence to collaborate with intelligent tools rather than feel dominated by them. Human oversight ensures that meaningful decisions remain with people, keeping algorithmic systems in an advisory role rather than an authoritarian one.

When these practices are in place, the researchers suggest, the corrosive link between AI adoption and psychological need satisfaction can be tempered. An organization that deploys AI transparently, involves its employees, invests in their understanding, and retains human judgment at critical junctures can capture efficiency gains without paying the psychological price. Conversely, an organization that rolls out intelligent systems abruptly, opaquely, and without worker voice may find that short-term productivity is purchased with long-term burnout, disengagement, and turnover—a trade-off that ultimately defeats the goal of sustainability.

The practical implications the authors draw for managers are concrete. They call for participatory AI design processes in which employees help shape the systems that will share their workplaces. They urge organizations to preserve employee discretion, deliberately carving out spaces where human judgment remains authoritative. They recommend systematic assessment of employees’ psychological health and wellbeing as part of AI adoption programs, treating mental health indicators with the same seriousness as operational performance dashboards. And they argue that AI adoption strategies should be synchronized with the United Nations Sustainable Development Goals—specifically SDG 3 on good health and wellbeing, SDG 8 on decent work and economic growth, and SDG 9 on industry, innovation, and infrastructure—so that technological transformation aligns with, rather than contradicts, broader sustainability commitments.

The study also situates itself within the fast-growing literature on responsible AI, but with a distinctive reorientation. Whereas much of that literature concentrates on fairness, bias, privacy, and accountability of algorithms, this paper insists that the human experience of working alongside AI deserves equal analytical weight. Moving from an efficiency-focused lens to a human sustainability lens, the authors contend, is not merely an ethical nicety but a strategic necessity for organizations that intend to thrive over decades rather than quarters. Their conceptual model is explicitly designed as a foundation for empirical testing, and the researchers invite scholars to validate the framework across different industries, organizational types, and cultural settings—particularly relevant given the international composition of the team, which spans Gulf, Middle Eastern, and South Asian research institutions.

As artificial intelligence accelerates into every corner of the service economy, the message of this research lands with urgency. The paradox it identifies—efficiency gains that quietly hollow out the human core of work—will not resolve itself. Organizations that treat AI adoption as a purely technical project risk discovering, too late, that their most valuable asset, a psychologically healthy and motivated workforce, has been degraded in the process. The study’s framework offers a way forward: embed transparency, participation, literacy, and human oversight into the fabric of AI deployment, and measure success not only in output but in the wellbeing of the people doing the work. In an era when the race to automate is fierce, the research suggests that the true finish line is not the smartest machine, but the most sustainable human-technology partnership.

Subject of Research: The sustainability paradox of artificial intelligence adoption and its effects on workers' psychological wellbeing and human sustainability

Article Title: Sustainability paradox of artificial intelligence adoption for workers’ wellbeing

Article References: Sustainability paradox of artificial intelligence adoption for workers’ wellbeing. (n.d.). https://doi.org/10.1007/s43621-026-04671-y

Image Credits: AI Generated

DOI: 10.1007/s43621-026-04671-y

Keywords: artificial intelligence, sustainability paradox, workers wellbeing, self-determination theory, recovery theory, AI literacy, human-centered AI, psychological need satisfaction, human sustainability, responsible AI, employee autonomy, organizational culture