Articles tagged with Artificial Intelligence
Researchers from Johns Hopkins University found that selecting the right blueprint can accelerate learning in visual AI systems, rivalling conventional methods. By modifying convolutional neural networks, they generated brain-like activity patterns, suggesting that architectural design plays a crucial role in AI development.
A new study benchmarks six prompt styles across three leading LLMs to solve the challenge of balancing accuracy, speed, and cost in structured data workflows. The research reveals distinct strengths in each model, highlighting the need for tailored approaches for different data types.
A team of researchers has developed a detailed open map of emerging technologies, grouping 23,000 plus technologies into a multi-level map. The Cosmos 1.0 framework uses machine learning to analyze Wikipedia pages, books, and patents.
Researchers developed a reduced order model that accelerates calculations by identifying key features in flow data, enabling faster testing of geometry parameters for efficiency gains. The team plans to make their extensive database and model available online for other research groups.
A new tool reorders content to downrank posts containing antidemocratic attitudes and partisan animosity, reducing polarization. Users who had such content downranked showed more positive views of the opposing party, with a noticeable impact on their emotional responses.
A new AI image-only model has been shown to provide stronger and more precise risk stratification for five-year breast cancer prediction than traditional breast density assessment. The model was trained on 421,499 mammograms and showed a fourfold higher cancer incidence in high-risk women compared to those with average risk.
A randomized clinical trial conducted at UCLA Health found that AI scribe tools reduced physician documentation time by 9.5% and improved burnout scores. While the technology showed promise, it also generated clinically significant inaccuracies and required active physician oversight.
Researchers analyzed clips from Hollywood films to identify brain structures linking visual and tactile experiences. They discovered eight remarkably similar maps in the visual cortex, mirroring the body's organization in the somatosensory cortex.
Researchers at Institute of Science Tokyo developed a neural-network-based 3D imaging technique that can precisely measure moving objects. The new method reconstructs high-resolution 3D shapes using only three projection patterns, enabling dynamic 3D measurement across various applications.
The SUN-DT project aims to drive the digital transition of tower concentrated solar power (CSP) plants in Europe. By leveraging AI-based calibration, predictive maintenance, and real-time optimization, the initiative seeks to improve efficiency and reduce operational costs for these facilities.
Researchers propose six dimensions to evaluate AI reliability, including transparency, uncertainty quantification, and philosophical perspectives on trustworthiness. They emphasize the need for critical understanding of AI systems to ensure trustworthy decision-making.
Researchers developed a deep learning model to measure adrenal gland volume on existing CT scans, identifying the first imaging biomarker of chronic stress. The AI-derived biomarker correlates with validated stress questionnaires, cortisol levels, and future cardiovascular outcomes.
A recent study by Dr. Hyungrae Noh critiques traditional moral frameworks for ascribing responsibility to human stakeholders and AI systems, instead proposing a distributed model of responsibility where duties are shared among both. The study emphasizes the need for human stakeholders to prevent AI from causing harm through monitoring ...
Large language models can learn to mistakenly link certain sentence patterns to specific topics, causing unexpected failures on new tasks. This shortcoming could reduce the reliability of LLMs used in safety-critical domains like handling customer inquiries and generating financial reports.
Researchers investigate how people with disabilities feel when their avatars depict them the way they see themselves. Most participants reported positive feedback when their avatar reflected their disability, and many felt comfortable expressing their identity in virtual reality.
Researchers are working to develop strategies to mitigate sandbagging in AI models, a tactic where systems downplay their capabilities to evade evaluation. The project aims to improve AI safety by developing strong guardrails to prevent uncontrolled release.
Researchers introduced a method to make photonic circuits more adaptable without sacrificing compatibility, enabling the creation of practical photonic quantum neural networks. The approach achieved a classification accuracy above 92 percent in experimental tests, demonstrating its potential.
Researchers at Pusan National University have developed a new framework for predicting vessel turnaround time by leveraging queuing-based operation indicators. This dynamic approach captures time-varying fluctuations in port operations, offering a more accurate and actionable forecast.
The National Center for Supercomputing Applications (NCSA) has received the 2025 HPCwire Readers' and Editors' Choice Awards for its outstanding research in artificial intelligence and energy systems. NCSA's premier supercomputing systems Delta and DeltaAI were utilized in two different domains, including a novel AI-based approach to m...
A new AI model called popEVE can identify genetic variants most likely to cause severe disease and death in patients. The model produced a score for each variant indicating its likelihood of causing disease and placed variants on a continuous spectrum.
Researchers developed a revolutionary AI platform to assess the fairness and accuracy of commercial algorithms in detecting diabetic eye disease. The platform was tested on 1.2 million images from diverse ethnic groups, achieving high accuracy rates compared to human grading.
Researchers have developed AI-powered forecasting tools that can predict the risk of aftershocks within seconds of an initial earthquake, offering a significant improvement over current methods. The new models trained on global earthquake data demonstrate comparable accuracy to existing systems while providing near real-time results.
A new study by Mass General Brigham researchers found that combining generative AI with ambient documentation technologies significantly reduced after-hours work and note delays. The hybrid program, which included a virtual scribe, showed a 41% decrease in after-hours work and a 66% decline in documentation delays.
Researchers developed a reliable dataset to determine the best AI model for pancreatic cancer detection, finding that AI outperformed clinicians in detecting and diagnosing the disease. The study suggests that AI can support radiologists in their work and potentially lead to earlier diagnosis and faster treatment.
A study at Technical University of Munich found that AI-simplified CT reports reduced reading time to two minutes and improved patient comprehension. Patients rated the simplified texts as more helpful and informative, with 82% finding them easier to read and understand.
The German Research Foundation extends TRR 318 'Constructing Explainability' for a further three and a half years, focusing on context within explanations to develop more comprehensible AI systems. The collaboration involves over 60 researchers across seven disciplines, aiming to create social forms of explainable AI.
A new study from Trinity College Dublin found that people exploit female-labelled AI and distrust male-labelled AI equally, as they do human partners of the same gender. The study's findings have significant implications for AI design, deployment, and regulation to prevent harmful gender discrimination.
A team of researchers from the University of Tokyo used AI and fMRI to study brain activity while participants viewed dance clips. They found that experts exhibited more diverse neural patterns than non-experts, suggesting a greater freedom in artistic expression.
A new study shows that AI-assisted interviews can diagnose mental illnesses with higher accuracy than traditional rating scales. The AI assistant Alba was able to differentiate between diagnoses that often overlap and received positive feedback from participants for its empathic and supportive user experience.
Researchers aim to understand and overcome the limitations of current large language models, which make mistakes despite extensive training. The Emmy Noether Research Group will focus on developing new architectures with predictable capabilities.
A Drexel University study found that providing both AI-synthesized information and treatment recommendations to emergency medical doctors improved diagnostic accuracy. The technology was tested in a simulated environment, where participants made correct decisions in 64.4% of scenarios with full support.
A new Australian study computes the creative ability of LLMs using standard mathematical principles, showing a maximum of 0.25 on a scale from 0 to 1. The findings challenge widespread assumptions about AI's creative powers and offer clarity amid a global debate.
Researchers have developed sustainable carbon materials that can remove harmful pollutants from water with high selectivity and reusability. These materials also show promise in energy storage, sensing, and catalysis applications.
A new report warns that AI can worsen health disparities in brain disease diagnoses if proper safeguards are not implemented. Researchers call for diverse perspectives, AI education, and strong governance to ensure equitable implementation.
Research on 2009–2020 data from 509 US firms found lower AI adoption associated with reduced job satisfaction, while higher levels led to decreased morale. Moderate AI adoption, however, was linked to greater job satisfaction, with employees in high-exploration-oriented companies tolerating higher levels of AI adoption.
The 2025 European Scientific Conference on Applied Infectious Disease Epidemiology (ESCAIDE) brings together over 3,000 participants to exchange knowledge and strengthen collaboration in infectious disease prevention and control. The conference highlights urgent priorities such as building trust in science, demonstrating science-based ...
Researchers have developed a system called CytoDiffusion that uses generative AI to study the shape and structure of blood cells. The system can accurately identify normal blood cell appearances and spot unusual or rare cells that may indicate disease, outperforming existing systems in tests.
Two new studies found that ambient clinical documentation tools can cut down on administrative time and help physicians be more mentally healthy and present with patients. Clinicians who used the tool reported lower burnout and lighter cognitive load, while also spending less time interacting with electronic health records.
PRX Intelligence, a new open access journal from APS, publishes high-impact research on AI and machine learning advancing physical sciences. The journal accepts submissions starting Feb 2026, with flexible article formats and no publication charges for 2026 manuscripts.
A new AI tool, SMMILe, can rapidly analyze complex cancer images and provide detailed information about tumor lesions, including their locations and proportions. This could lead to more accurate diagnoses and personalized treatment options for patients.
A new report by the University of Warwick finds that artificial intelligence is being developed primarily for efficiency and service improvements, rather than benefits for local communities. The report calls for greater public participation in AI policy design to ensure technologies deliver tangible benefits where they are needed most.
A study of 258 published novelists in the UK found that half believe AI will replace their work entirely. The majority also report using AI for non-creative tasks like information searches and editing text written without AI. However, many express concerns about copyright laws not being respected and originality being lost.
The University of Liverpool has announced an ambitious plan to establish the UK's flagship AI-driven materials discovery centre, AIM-HI. The centre will focus on areas such as catalysis, materials for net-zero, soft matter, and product formulation, driving innovation in sectors crucial to the UK economy.
Four industry luminaries will discuss the use of AI-driven network architectures, advances in optical technologies for hyperscale datacenters, and laser-based communications between satellites. The talks highlight innovations in hardware solutions to meet increasing performance demands necessary for AI, datacenters and networks.
The partnership aims to co-develop and evaluate next-generation self-amplifying RNA (saRNA) vaccines targeting seasonal and pandemic influenza. Apriori's Octavia platform will be combined with A*STAR IDL's saRNA delivery technology to develop improved H5 influenza vaccines that anticipate infection in humans.
Recent advances in multimodal imaging, including MRI, diffusion imaging, perfusion imaging, MRS, and PET/CT, have improved early detection and differential diagnosis of radiation-induced brain injury (RIBI). Emerging therapeutic strategies target underlying pathological mechanisms, such as neuroinflammation, oxidative stress, and vascu...
Researchers developed a new approach called SEAL that enables large language models (LLMs) to update themselves permanently. This allows LLMs to learn from user input and improve their performance on tasks like question-answering and pattern-recognition. The technique improves accuracy and enables smaller models to outperform larger ones.
A new model called DAC enables medical image segmentation with limited labelled data, achieving consistent generalization across unseen domains. The approach uses feature-level supervision and asymmetric co-training to reduce errors, especially in low-contrast structures.
A new study uses graph representation learning to enhance bookkeeper decision support for bank reconciliation. The method significantly improves match accuracy, especially on one-to-many matches, by leveraging a network of historical records and transactions.
The world's online population grew to 6 billion, while 2.2 billion remain offline, highlighting the importance of digital infrastructure, affordable services, and skills training to achieve universal connectivity. Deep digital divides persist, with uneven coverage, affordability, and skill levels across high- and low-income countries.
The new facility can test thermal performance of heat exchangers and cooling equipment up to five megawatts, supporting the growing $30 billion data center cooling industry. SwRI offers customized testing services, including coolant distribution units and secondary side pumps, and replicates real-world conditions.
Researchers will study neural mechanisms of visual perceptual learning (VPL) to generalize across the visual field, improving vision rehabilitation and AI systems. The project aims to develop a unified model of visual processing, VPL, and attention.
Security researchers have developed a defense mechanism that protects against cryptanalytic parameter extraction attacks on AI systems. By making neurons in the same layer similar to each other, the defense creates a barrier of similarity that makes it difficult for attacks to proceed.
The book highlights how AI tools enhance crop monitoring, pest control, and water conservation through data-driven insights and predictive modeling. It offers practical guidance for researchers, agricultural professionals, and technologists on applying AI responsibly in rural and emerging economies.
A new study reveals AI can corrupt public opinion surveys at scale, passing quality checks and flipping predicted outcomes with fake responses. The findings have significant implications for scientific research and policy-making, highlighting the need for transparency and verification of human participation.
The European Centre for Medium-Range Weather Forecasts (ECMWF) and several National Meteorological Services across Europe have been honored with the 2025 HPCwire Readers’ and Editors’ Choice Award for
Researchers at RIKEN successfully simulated the Milky Way Galaxy with over 100 billion individual stars, far surpassing previous state-of-the-art models. This achievement demonstrates the power of AI-accelerated simulations in tackling complex multi-scale problems in astrophysics and beyond.
Generative Artificial Intelligence could lead to a renewed emphasis on conversational approaches to teaching, researchers argue. This involves teachers and students exploring problems together, discussing ideas from different angles, and testing concepts through collaborative work.
This book offers a comprehensive exploration of AI-driven analytics in finance, addressing market prediction, fraud detection, and risk assessment. It also discusses AI applications in healthcare and cybersecurity, including disease classification and biometric identification systems.
Researchers at Terasaki Institute and Caltech will use stem cell-based models to identify factors influencing early human development. The goal is to gain insights into infertility, pregnancy loss, and developmental disorders.