generative ai

Complete Defectivity and Simplex-Range Rational Identifiability for Homoscedastic Gaussian Moment Varieties

This work gives a complete dimension and defectivity classification for homoscedastic Gaussian moment varieties over an algebraically closed field of characteristic zero. It proves that these varieties have the expected dimension for every moment order at least four and combines this result with the known cubic classification to determine all defective cases. The paper also establishes a uniform nondefectivity theorem when the common covariance is restricted to a general positive-dimensional linear subspace, including the isotropic covariance model. In addition, it determines rational identifiability throughout the simplex range. The least rationally identifying moment order is two for one component, five for two components, and four for every model with at least three components in the simplex range. The proofs combine a degeneration of the common covariance tangent block, fat-point postulation, Waring decomposition methods, cumulant coordinates, and flat moment matrices. The three- and four-component cases are treated separately by explicit rational reconstruction and an exact reduced Gröbner-fiber argument with boundary exclusion. The accompanying computation archive provides executable exact-arithmetic verification code, fixed inputs, deterministic outputs, and a unified reproduction command. It includes independent checks of Jacobian ranks, the cubic defect classification, quartic recovery, restricted covariance models, the numerical conditions entering the fat-point argument, and the exact small-component fiber certificates. Research methodology and AI assistance:This work was developed using the CARMA-Math research workflow, a cumulative AI-assisted mathematical research methodology using persistent research archives, literature and prior-art investigation, iterative proof exploration, and verification procedures. Generative AI (ChatGPT) was used extensively for mathematical exploration, proof development, computational reasoning, literature research, and manuscript preparation.

MERFISH MultiCSV DotPlot

MERFISH MultiCSV DotPlot is a lightweight Python utility for generating dot-plot style summary figures from multiple MERFISH cell-by-gene CSV datasets and multiple selected genes. Each input CSV is represented as one column and each selected gene as one row. Dot size represents the fraction of cells with expression greater than a user-defined positive-cell threshold. Dot color can represent either absolute mean expression or per-gene relative mean expression scaled from 0 to 1 across the supplied datasets. The software accepts an arbitrary number of MERFISH CSV files and an arbitrary number of genes. It exports both publication-style figures and a long-format numerical summary table. Two color modes are available:- "absolute": dot color represents mean expression on the original input scale.- "gene_scaled": mean expression is independently scaled from 0 to 1 for each gene across the supplied datasets, emphasizing relative expression patterns between datasets. Positive cells are defined as:expression > positive_threshold The positive-cell threshold is explicitly specified by the user because appropriate thresholds depend on the expression scale and preprocessing of the input data. No automatic log/linear transformation is performed. Expression values are used exactly as supplied in the input CSV files.Input CSV format: Each CSV should contain one cell per row and gene-expression values in gene-named columns. Additional metadata or coordinate columns (e.g., cell ID, x, y, z, section information) may also be included. All genes selected for plotting must be present as numeric columns in every input CSV. The software was functionally validated using MERFISH-derived cell-by-gene datasets with multiple cell populations and genes. Expected qualitative differences between canonical D1- and D2-associated gene-expression patterns were reproduced during validation. No third-party MERFISH dataset is distributed with this software. Generative AI (ChatGPT, OpenAI) was used to assist with code generation, refinement, testing, packaging, and documentation. The concept, intended scientific use, validation, and final responsibility for the software remain with the author.

Complete Defectivity and Simplex-Range Rational Identifiability for Homoscedastic Gaussian Moment Varieties

This work gives a complete dimension and defectivity classification for homoscedastic Gaussian moment varieties over an algebraically closed field of characteristic zero. It proves that these varieties have the expected dimension for every moment order at least four and combines this result with the known cubic classification to determine all defective cases. The paper also establishes a uniform nondefectivity theorem when the common covariance is restricted to a general positive-dimensional linear subspace, including the isotropic covariance model. In addition, it determines rational identifiability throughout the simplex range. The least rationally identifying moment order is two for one component, five for two components, and four for every model with at least three components in the simplex range. The proofs combine a degeneration of the common covariance tangent block, fat-point postulation, Waring decomposition methods, cumulant coordinates, and flat moment matrices. The three- and four-component cases are treated separately by explicit rational reconstruction and an exact reduced Gröbner-fiber argument with boundary exclusion. The accompanying computation archive provides executable exact-arithmetic verification code, fixed inputs, deterministic outputs, and a unified reproduction command. It includes independent checks of Jacobian ranks, the cubic defect classification, quartic recovery, restricted covariance models, the numerical conditions entering the fat-point argument, and the exact small-component fiber certificates. Research methodology and AI assistance:This work was developed using the CARMA-Math research workflow, a cumulative AI-assisted mathematical research methodology using persistent research archives, literature and prior-art investigation, iterative proof exploration, and verification procedures. Generative AI (ChatGPT) was used extensively for mathematical exploration, proof development, computational reasoning, literature research, and manuscript preparation.

Digital Financial Inclusion and Its Impact on Economic Growth

The rapid growth of Generative Artificial Intelligence (GenAI) has made AI literacy an essential skill for students, yet conventional classroom and e-learning platforms provide limited support for helping learners understand, question, and critically evaluate AI-generated information. This paper presents an AI-Driven Instructional Support System with Dynamic Learning Analytics for Classroom Evaluation, a role-based, web-based educational platform branded EduMentor that integrates academic and assessment management with a Generative AI instructional-support layer. The system employs Retrieval-Augmented Generation (RAG) and web-scraping-based verification to ground AI responses in course material and trusted external sources, assigns a confidence score to each generated response, and classifies learning content using Bloom's Taxonomy to support progressive cognitive development from Remember and Understand to Apply, Analyze, Evaluate, and Create. A Django backend, a relational database, and a Flutter-based client provide role-specific interfaces for Administrator, HOD, Staff/Teacher, and Student users, along with assessment management, student-performance analysis, and a learning-analytics dashboard. The system was evaluated using 100 functional test cases spanning authentication, assessment management, performance analysis, AI-generated recommendations, and application communication. The platform achieved 94% overall functional accuracy, 100% authentication accuracy, 89% AI-recommendation relevance accuracy, a usability score of 4.4/5 (88%), and an average API response time of 1.8 seconds. The results indicate that an integrated role-based platform combining assessment data with AI-assisted instructional feedback is a feasible and practical approach for data-driven classroom evaluation, provided that AI-generated recommendations continue to be treated as decision-support output requiring teacher verification.

Artificial Intelligence and Future-Oriented Aesthetic Education: Design Thinking in Middle School Art Curriculum

Middle school art curricula in many systems still center on skill reproduction, which sits uneasily with the rapid diffusion of generative AI. This study develops a conceptual instructional model that embeds generative AI within the five-stage design thinking process (empathize, define, ideate, prototype, test) for an eighth-grade visual communication course. AI is assigned differentiated functions across the stages—analytical, retrieval, generative, assistive, and evaluative—while students retain responsibility for aesthetic judgment, justification of selections, and iterative revision. Three learning outcomes anchor the design and are operationalized in parallel measurement instruments: aesthetic judgment, creative problem-solving in visual communication, and stylistic self-awareness. Communicative intent is treated as the core sub-dimension of creative problem-solving rather than as a separate outcome, since intent is observable only through the design products and decisions that enact it. A quasi-experimental pretest-posttest pilot was conducted with 68 eighth-grade students (experimental group n = 34, control group n = 34) over a ten-week intervention at a single school, with both classes taught by the same art teacher. ANCOVA indicated that the experimental group outperformed the control group on aesthetic judgment, F(1, 65) = 8.42, p = .005, ηp² = .115, and creative problem-solving, F(1, 65) = 9.17, p = .003, ηp² = .124, with smaller yet significant gains on stylistic self-awareness. Thematic analysis of journals and interviews surfaced three patterns: shifted attention from execution to selection, prompt-based reasoning, and unease about authorship. Findings are presented as encouraging pilot evidence under specific local conditions, pending replication beyond the present site.

AI-driven drug discovery and personalized medicine: integrating quantum computing and next-generation technologies

Generation and Perception: A Computational Evaluation Method for Visual Quality and Emotional Impact in AI Artworks

Key contributions include proposing a multi-task learning framework for jointly optimizing visual quality and emotion, establishing the inaugural VAWE-Art dataset comprising 5,000 AI-generated images with 20-dimensional emotional annotations, and providing computational foundations for emotion-controllable generative art systems.

Bland new world: is AI making us all think the same?

Engineering student perceptions of generative AI use in learning trust judgment and information seeking

Generative AI is rapidly reshaping higher education, yet its influence on student learning in engineering remains insufficiently understood. In courses that require conceptual understanding, design reasoning, and problem-solving, AI tools may support learning by providing explanations, solution pathways, and feedback. At the same time, they raise questions about trust, verification, and the quality of student thinking. This study investigates student perceptions of generative AI in engineering education, focusing on learning, trust, and information-seeking. This work extends a prior conceptual framework grounded in Ellis’s theory of information-seeking. The study applies a six-stage subset of Ellis’s model—starting, chaining, browsing, differentiating, monitoring, and verifying—to examine how students perceived generative AI during an engineering learning activity. It also examines themes related to perceived usefulness, learning support, trust, judgment, and AI use. Within this sample, respondents generally reported perceiving AI as a support tool for conceptual understanding, efficiency, and checking their reasoning. Among respondents who reported at least some AI use, higher agreement was observed for items concerning ongoing task support, perceived deeper learning, and the application of their own judgment, whereas lower agreement was observed for using AI to understand the problem requirements initially or to check and validate technical work.

Large language medicine: defining a new paradigm in human health

Artificial intelligence (AI) is poised to revolutionize our understanding of disease and pre-disease, which could transform the way medicine is practiced. Advances in deep learning systems, alongside the advent of generative AI and large language models, promise to usher in an era of multimodal AI systems that permeate every aspect of medicine. AI is expected to support the full spectrum of public health and clinical functions, including surveillance, monitoring, protection, health promotion, and disease prevention. In parallel, molecular biology has been revolutionized by AI solutions that can accurately model proteins and other biological molecules at scale with the potential to accelerate drug discovery and reshape our understanding of disease and pre-disease mechanisms. However, AI has yet to be incorporated into common day-to-day practice, underscoring the difficulty of clinical integration. Achieving this goal will require advancing algorithmic architecture, sourcing higher-quality multimodal data, increasing processing power and efficiency, developing secure and fit-for-purpose data infrastructures, ensuring interoperability with diverse health systems and clinical workflows, and establishing regulatory pathways to protect patient safety and clarify clinician liability. The potential emergence of autonomous systems capable of carrying out an increasing number of clinical tasks raises important ethical and legal questions about accountability and responsibility in patient care. This review explores the current state of AI in medicine, highlighting that current clinically mature and regulatory-approved products are based on deep learning architecture and are predominantly diagnostic. Robust regulatory frameworks and ethical guidelines must be established to govern the development and deployment of AI, ensuring alignment with patient safety, clinical guidelines, and public trust. Multidisciplinary collaboration among clinicians, researchers, technologists, ethicists, regulators, and policymakers is essential moving forward.

Is open, distance, and digital education (ODDE) good for the environment? A systematic review of carbon footprint studies

This study presents a systematic review of research on the environmental sustainability of Open, Distance, and Digital Education (ODDE), focusing on its carbon footprint. Although higher education institutions (HEIs) are expected to reduce greenhouse gas emissions, the environmental impacts of teaching and learning modes remain underexplored compared to campus operations. To address this gap, 17 studies published between 2002 and 2025 were examined. The evidence shows that ODDE typically generates far lower emissions than campus-based education, mainly through reduced travel and campus energy use and the benefits of centralized delivery. Landmark studies reported reductions of more than 80% compared to face-to-face teaching. Studies conducted during the COVID-19 pandemic confirmed significant savings but also highlighted rebound effects, as household energy use increased while campus facilities continued to consume resources. Emerging technologies, particularly generative AI, further complicate the picture due to their high energy demands. Overall, the evidence suggests that ODDE can contribute positively to reducing the carbon footprint of higher education. However, the small number of studies, lack of standardized methodologies for carbon footprint assessment (CFA), and limited attention to rebound effects constrain broader generalizability. Future research should systematically compare different delivery models across diverse institutional contexts and develop shared CFA guidelines. Strengthening the sustainability profile of ODDE is of critical importance, not only for the higher education sector but also for advancing global climate goals.

Do Design Keywords Matter for Generative AI? An Empirical Study of a Design Element-Based Prompt Framework