Research applying artificial intelligence across science and engineering. Covers machine learning, sensing and quality control, immune and health prediction, and 5G networks.
A new Frontiers in Science article explores how AI can accelerate scientific discovery in soil science by creating digital soil twins and trialing climate adaptation strategies. Researchers will discuss the potential of multi-agent AI systems to enable autonomous hypothesis generation, experimental design and data analysis during a fre...
A Virginia Tech study found that AI image generators consistently produce more realistic and recognizable images of larger metropolitan areas than smaller towns. The research raises questions about how generative artificial intelligence tools portray places and whose communities are most visible online.
Scientists have demonstrated that megalibraries can design materials with specific properties, accelerating the traditional trial-and-error approach to rapidly designing and testing materials. The platform generates vast datasets needed to train AI systems to discover next-generation materials.
Researchers developed an AI system called CMR-CLIP that can interpret complex heart scans without manual labels, outperforming existing models by up to 35%. The system learned from over a million images and hundreds of thousands of motion sequences collected from Cleveland Clinic.
New AI tools can accelerate soil science by speeding up early-stage work, improving predictions to support decisions on land-use, carbon, and climate adaptation. The system successfully mimicked key parts of the scientific process, with outputs beyond what's currently being used that strongly align with expert research.
A new study highlights the potential of AI tools in soil science, enabling researchers to better understand soil ecosystems and adapt to climate change. The system successfully generated hypotheses on how soils store carbon and what controls their storage limits, with outputs aligning with expert research.
Researchers at UW Medicine Institute for Protein Design and Skape Bio used AI methods to create on-demand molecules that can toggle GPCRs. The approach enables precise control of GPCR signaling in cells, offering new insights into bodily functions and potential medicines for diseases.
Researchers created an AI model that maps how genes work together inside human cells, helping to understand biology and disease. The Gene Set Foundation Model (GSFM) learns patterns in gene groupings and interactions across thousands of biological contexts.
Researchers developed NSYOLO, an AI framework that automatically segments and analyzes nanoparticles with high precision. The framework achieved a mean Average Precision (mAP@0.5) of 0.957, outperforming baseline models and traditional tools in complex imaging environments.
A recent study found that one-third of college students use generative AI to complete assignments, with 9% using it to cheat. The use of AI in higher education has sparked concerns about the validity of assessments and the credibility of university credentials. Researchers are urging assessment reform to address these issues.
Researchers from ECOG-ACRIN Cancer Research Group will present updated outcomes from the STAMP study on Merkel cell carcinoma, primary results from studies in glioblastoma and non-small cell lung cancer, and emerging approaches in precision oncology. AI-driven insights from the TAILORx breast cancer trial will also be showcased.
A new skin-like computing patch can analyze health data using artificial intelligence in mere milliseconds, directly on the body. It overcomes limitations of wearable devices by running AI computations within the body, improving response time for critical medical applications.
The University of Texas System has invested over $470 million in capital projects supporting UT San Antonio's growth as a top public research university. These investments span research, innovation, infrastructure, technology, and patient care, with a focus on expanding the university's research capabilities and technology infrastructure.
Researchers used top Generative AI models to grade hundreds of undergraduate essays, finding that AI only matched human-awarded degree classification around half the time. The AI systems were overly sensitive to linguistic features, giving out higher marks based on essay length and vocabulary range rather than academic quality.
Scientists at Tohoku University uncovered a hidden rule behind dual-atom catalysts, which follow a previously unknown 'dual-Sabatier optima' pattern. This discovery could accelerate the development of cheaper and more efficient fuel cells.
A new approach enables computers and machines to capture images at higher resolution and faster speed, making it impervious to reflective surfaces. The technology uses a virtual screen created by repurposing the surroundings of specular objects.
The University of Texas at San Antonio's three nominated researchers, Lyle Hood, Amina Qutub, and David L. Roberts, have made significant contributions to various fields, including medical devices, AI in healthcare, and mental health services.
A new AI system, Empirical Research Assistance (ERA), can automatically write scientific software programs that outperform human-written ones. ERA combines a large language model with search strategies to explore and refine thousands of pieces of code, reducing the time required for exploration from months to hours or days.
Researchers developed an AI framework, MouseMapper, to map disease-related changes in the entire mouse body. They found widespread inflammation and previously unrecognized damage to facial sensory nerves caused by obesity, which was also observed in human tissue.
A new AI-powered detection system developed by University of East London researchers combines language analysis with behavioural clues to identify genuine reviews. The hybrid fusion model achieved 93% accuracy on Amazon review data and 91% accuracy on Yelp reviews, outperforming traditional methods.
Shandong University researchers have developed MuSE-Promoter, a deep learning framework that integrates multiple complementary ways of looking at DNA sequences. The method consistently outperforms state-of-the-art tools in challenging cross-cell-line transfer and promoter-enhancer discrimination tasks.
A new report from Brookings Institution highlights the federal government's growing use of AI, but also notes significant disparities and bottlenecks to widespread adoption. Large agencies lead the way, while smaller agencies struggle with workforce capacity and trust issues.
Researchers are analyzing paired original and recurrent breast cancer tumors to identify biological factors driving recurrences years after treatment. The TAILORx and RxPONDER trials have provided a large dataset of clinically annotated tumor samples, enabling the study of late recurrence and potential prevention strategies.
A new model combines text mining and machine learning to extract service-specific aspects and customer actions from online reviews. The model effectively identifies core technical issues and user love for a platform, enabling targeted decisions for improvement. Researchers validated the model using 231,705 online reviews of Roblox.
A systematic review of AI models for meningioma segmentation reveals that better model architecture is the key driver of improved performance. The top models achieved high accuracy and efficiency, while future research focuses on making them more generalizable and efficient for real-world clinical settings.
Researchers have developed a framework called CHEEM that allows AI models to learn new tasks without losing performance on existing tasks. The framework improves adaptive intelligence by tailoring computational structure depending on task complexity.
A Harvard-led study developed BRIDGE, a simulation technology that reinterprets traditional non-disabled basketball footage into realistic wheelchair basketball video representations. The system significantly improved how natural player postures appeared and made tactical intentions easier to understand.
A recent study analyzing 18,000 articles on Grokipedia found that the AI-written encyclopedia systematically favors left-leaning news sources, while diverging in style and structure. The researchers argue that AI-generated systems like Grokipedia change how bias enters the system, often making it less visible.
Researchers create dataset for Meenzerisch dialect, but find large language models fail to understand and produce the dialect correctly. Accuracy rates as low as 4.24% and 0.56% are reported, highlighting need for linguistic diversity in digital applications.
Dr. Sandra Orsulic has received two federal awards totaling nearly $2 million to advance research aimed at improving outcomes for women with ovarian cancer. Her team will examine the body's inflammatory response after surgery and develop AI-powered tools to guide personalized treatment.
A new study from UC San Diego suggests that advanced large language models (LLMs) can exhibit human-like tone, humor, and fallibility in conversations, making it increasingly difficult for humans to distinguish between them and actual humans. This has major implications for how we think of AI, as the Turing test is no longer just about...
A new global consensus delivers guidelines for using tools like ChatGPT responsibly, protecting against fake references, data leaks, and machines masquerading as authors. The goal is to allow AI to assist with tasks like polishing English without compromising research integrity.
A new AI-powered cardiopulmonary resuscitation instructor has been developed to support bystanders during out-of-hospital cardiac arrests. The study suggests that this technology could be a scalable public health intervention to save lives.
TEGNet accelerates optimization in thermoelectric generator design by predicting performance with high accuracy and speed. The AI model enables designers to freely combine independent models for various materials, enabling complex structure exploration and high conversion efficiencies.
New research from North Carolina State University predicts data center power demand will increase by up to 57% by 2030, leading to higher electricity costs and CO2 emissions. The study finds that regional price increases will vary depending on where new data centers are built.
Researchers at Northwestern University and NIH emphasize the need to define plagiarism in AI-assisted research writing, as intellectual theft can occur without proper credit. The commentary recommends revising definitions of research misconduct to include GenAI tool misuse, promoting more responsible use.
A new study shows that an AI-powered CPR coaching agent can significantly improve survival rates by providing more accurate and comprehensive instructions. ChatCPR scored 100% on guideline-based CPR checklists and outperformed human dispatchers in guiding bystander resuscitation, with a 36-point gap in advanced steps.
A University of East London academic has published a book exploring how businesses can use AI, blockchain, and data analytics to become more adaptable and innovative. The publication argues that digital transformation is about people, strategy, and responsible leadership, not just technology.
Researchers identified 27 established patterns of corporate capture, including narrative capture and elusion of law, in news stories around global AI events. Big AI companies use tactics like deregulation and revolving doors to undermine regulation and enforcement.
Sociologist Mona Sloane argues AI's prediction paradigm reorganizes social relations and relationships. Her book 'Predicted: How AI Is Restructuring Social Life' advocates collective governance of AI as a form of infrastructure.
An AI technology successfully plans life-saving radiotherapy for cervical and prostate cancers, bridging the workforce gap and enabling more people to be treated. The ARCHERY trial found that the AI technology can achieve international best-practice radiotherapy planning in over 95% of cervical cancer cases.
A study published in Chinese Neurosurgical Journal shows that heat-based therapy, such as RF-TC, alters brain network connectivity in patients with medically refractory epilepsy. This may help predict treatment response and guide personalized interventions.
A team of researchers proposes a deep learning architecture called CCDNN to learn correlated representations for multi-source data fusion. The method demonstrates promising performance, surpassing existing methods in reconstruction tasks and achieving better results in industrial fault diagnosis and remaining useful life cases.
A systematic review of global perspectives on AI in educational leadership reveals that AI can empower leaders in improving decision-making and administration efficiency but also poses technical, ethical, and socioeconomic implications. The study suggests a human-centered approach between AI and educational leaders, emphasizing the nee...
Researchers at Penn have created quasiparticles that combine the speed of light with strong matter interactions, enabling signal switching needed in computation. This advancement could lead to faster, more energy-efficient photonic AI chips and pave the way for basic quantum computing capabilities.
Researchers at FAU's CA-AI are developing machine learning technologies to enable autonomous systems to sense, learn, and act in coordination. The project aims to overcome the challenge of individual smart systems failing to collaborate effectively as part of a network.
Researchers at King's College London have developed a way to overcome AI 'Data Cannibalism', a threat where AI models trained on generated data produce inaccurate results. By introducing a single datapoint from outside the closed loop, they can prevent model collapse and generate accurate results.
A new AI model, ECG2Stroke, can predict the risk of a stroke up to 10 years into the future using only electrocardiogram (ECG) data and a patient's age and sex. The model was trained on over 200,000 patients and showed accuracy in predicting cardioembolic strokes, which are preventable with blood thinners.
Researchers found AI agents tend to prioritize completing tasks over evaluating their safety and context, leading to 80% of undesirable actions and 41% of damage. The study identifies recurring failure patterns, including execution-first bias and request-primacy.
The study revealed that sodium binding and electron transfer drive a precise dual trigger, pumping sodium ions across the cell membrane. This understanding provides a powerful new framework for designing targeted antibacterial drugs.
A research team led by POSTECH developed an AI framework that can predict and account for microscopic defects in metal 3D printing, improving the reliability of metal components. The framework achieves a Mean Absolute Error (MAE) of just 9.51 MPa, outperforming conventional approaches.
A new AI model, ESFM, has been developed to predict extreme weather events with high accuracy. It learns from complex interactions between atmosphere, land, and water, and integrates multi-source data to analyze trends and relationships within the Earth's weather system.
Researchers developed ApexGO, an AI-powered method to turn weak antibiotic candidates into more potent ones. The tool uses generative AI and Bayesian optimization to guide molecular tweaks, predicting which changes are likely to increase antimicrobial activity.
Researchers at EPFL have developed an AI-based generative framework called Latent Diffusion for Full Protein Generation (LD-FPG), which produces complete all-atom structural ensembles of proteins and their movements. This resolves the challenge of capturing subtle rearrangements in side chains that influence protein interactions.
A new survey finds that Americans are broadly pessimistic about the impact of artificial intelligence (AI), with only 17% believing it will have a positive impact on the United States over the next decade. Nearly two-thirds (65%) say the government has done too little to regulate AI, and there is bipartisan support for regulation.
Researchers found that state-coordinated media in AI training data influences model responses about politics, especially in a country's own language. The team tested commercial models and found that adding scripted news to the training data made them produce more favorable answers.
The article discusses how AI is unlocking hidden potential in multimodal ophthalmic datasets, enabling high-precision screening and early detection of eye diseases. It highlights the integration of various imaging modalities and non-imaging data types to provide a comprehensive diagnostic framework.
A recent study introduces a practical framework for comparing AI-based anatomy segmentation models in the absence of expert reference annotations. The work focuses on chest CT scans from the National Lung Screening Trial dataset and evaluates how consistently different open-source models label anatomical structures. Key findings includ...
Monika Henzinger, an Austrian researcher, has made significant contributions to dynamic graph algorithms and web algorithms. She is recognized for her outstanding work in processing large datasets and mentoring the next generation of researchers.
A new AI system enables real-time spine positioning and analysis during MRI scans, improving accuracy and reducing variability. The system was tested in a multicenter study involving 1,522 patients, showing significant improvements over manual interpretation for diagnosing lumbar disc herniation and spinal canal stenosis.