Articles tagged with Artificial Intelligence
A systematic review of 53 studies found that machine learning algorithms for predicting suicidal behavior have low accuracy. The algorithms were poor at identifying individuals who would later self-harm or die by suicide, while exceling in detecting those who would not.
Researchers used AI to analyze protein complexes, discovering catch-bonds 'switch on' almost immediately after force is applied. This finding has implications for understanding bacterial attachment, tissue resilience and developing new biomaterials.
Researchers developed AI-assisted conductive hydrogel dressings for intelligent wound monitoring, combining real-time physiological signal detection with artificial intelligence. The smart dressings offer personalized wound care and have the potential to bring hope to millions suffering from chronic wounds.
Scientists have discovered key circuit changes that lead to the maturation of reliable brain activity patterns in response to visual experience. The findings suggest that the brain develops a modular organization, with patches of neurons activating together in response to specific features, allowing for efficient learning and adaptation.
A new University of South Florida study highlights the impact of menopause on women's voices, particularly those who rely on their voices professionally. The research outlines a range of treatment options, including AI-powered voice biomarkers, to detect subtle vocal changes linked to menopause.
Bobin Deng's research focuses on developing AI tools that can run directly on mobile phones, drones, and other smaller systems without relying on costly supercomputers. His technique uses activation sparsity to predict which data will be activated, reducing memory usage and energy demands.
A new study proposes a unified framework for understanding risk as a result of human cognition, connecting perception and analysis. The model highlights the disparity between target value expectations and realistic ones, emphasizing the role of self-aware stakeholders in transforming perception into rational response decisions.
The Leopoldina Annual Assembly will focus on the progress of artificial intelligence in various fields, including medicine, image creation, medical diagnoses, and decision-making processes. Experts from natural and social sciences will discuss the effects of AI on society, ethics, and democracy.
Experts weighed in on the current state of artificial intelligence (AI) in cancer care, noting both its potential for improving efficiency and its challenges. Many expressed excitement over AI's possibilities for helping support an overburdened oncology workforce and accelerating the pursuit of new cures.
A new way to fight drug-resistant infections has been discovered through the study of 'microbial piracy', where bacteriophages hijack other viruses to break into bacterial cells and spread. This could potentially be harnessed for medicine, offering a major route to tackling antimicrobial resistance.
A recent study using AI and satellite images estimates that fewer than 600,000 wildebeest migrate across the Serengeti-Mara ecosystem annually, contradicting previous aerial surveys. The findings highlight the potential of satellite-based population counts for terrestrial mammals.
Researchers at New York University have created a framework to control swarm intelligence in AI, using geometric design rules to mimic natural behaviors like flocking birds and schooling fish. The framework enables robots to curve and interact with each other based on 'curvity,' a quantity that drives collective behavior.
A new AI-powered satellite survey reveals a significant shortfall in estimated Serengeti wildebeest numbers, challenging traditional aerial surveys. The study's findings suggest at least 700,000 fewer wildebeest than previously estimated, highlighting the need for accurate data to inform conservation efforts.
A team of researchers developed an AI-powered tool, called the Virtual Research Assistant (VRA), to filter through thousands of data alerts and identify genuine signals caused by supernovae. The VRA successfully filtered over 30,000 alerts while missing fewer than 0.08% of real supernovae alerts.
A new AI-driven AFM platform accurately identifies macrophage polarization states and distinguishes between M0, M1, and M2 phenotypes. The model successfully validated using flow cytometry data, showing promise for disease diagnostics and therapeutic responses.
Researchers have developed a silicon chip that uses light to perform convolution operations for AI, reducing energy consumption and increasing speed. The chip achieves near zero energy performance, leap forward for future AI systems.
Researchers harness AlphaFold 3 to predict how T cells recognize peptides, opening avenues for precision immunotherapy and vaccine design. The approach enables in silico identification of immunogenic epitopes that could serve as vaccine targets.
Researchers have developed a new light-based chip that cuts power consumption for image recognition tasks by up to 100 times, using lasers and microscopic lenses fabricated onto circuit boards. This breakthrough enables faster performance and potentially strain-free AI systems.
A team of researchers has created a reliable method for accurately recording greenhouse gas emissions from corporate sustainability reports. The dataset, published in Scientific Data, provides a rigorous and transparent foundation for evaluating automated approaches to sustainability reporting.
Researchers developed an AI tool to assess tambaqui fish stress levels based on their lower half body coloration. The study found that stressed fish became darker, and the trait is heritable, impacting animal welfare and farming practices.
Researchers used AI models with integrated clinical and claims data to predict chronic kidney disease (CKD) progression to end-stage renal disease (ESRD). The study found that the models outperformed single data source models and reduced racial bias. The findings can inform likelihood and management of CKD, supporting targeted interven...
The International Telecommunication Union is strengthening its collaboration with academia to gather insights on emerging technology trends. The new Academic Advisory Body will produce informative policy briefs and forward-looking analyses on the evolving role and impact of innovation.
The Global RETFound initiative aims to develop the world's first globally representative AI foundation model in medicine, addressing health inequalities and AI bias. The project involves a diverse dataset of over 100 million fundus photographs sourced from 65+ countries.
A new AI oracle aimed at solving complex physics problems has received $19.4 million in funding from the US Department of Energy. The project, led by the University of Michigan, will focus on developing artificial intelligence models to generate trustworthy data for engineering design of complex devices.
A new study from the University of Turku found that AI systems can accurately evaluate social features from images and videos, comparable to human evaluations. The researchers compared AI's evaluations with those made by over 2,000 humans, showing that AI's results were consistently accurate.
Researchers at UMC Utrecht developed a new AI-powered printer called GRACE that can print implantable tissues with improved cell survival and functionality. The printer uses computer vision and laser-based imaging to design and print complex structures, including blood vessels and cartilage layers.
A recent breakthrough in MXene nanomaterials could enable bespoke design tailored by AI. Researchers have discovered the thermodynamic forces underlying their unique structure and behavior, setting parameters necessary for AI-guided design.
Researchers used AI to identify four times more earthquakes than earlier tools and pinpoint previously unknown faults in the region. The study expands seismicity recorded by monitoring stations from 2022 to 2025, revealing two faults converging under the town of Pozzuoli west of Naples.
Researchers at UCR have developed a method to preserve AI safeguards in open-source models by retraining internal structure to detect and block dangerous prompts. The approach avoids external filters or software patches, instead changing the model's fundamental understanding of risky content.
Researchers developed a powerful new AI algorithm called RoboBallet that enables large sets of robotic arms to work together faster and smarter. The system plans movements and tasks automatically, unlocking new levels of flexibility and efficiency in manufacturing.
A new method, AEquity, identifies and corrects bias in health care datasets used to train machine-learning algorithms. The tool can spot well-known and previously overlooked biases across different datasets.
The study found that humans and AI systems interact with flexible and incremental learning modes similarly to working memory and long-term memory in humans. Quicker, flexible in-context learning emerges after a certain amount of incremental learning has taken place, similar to human learning.
The use of AI in the criminal justice system raises concerns about fairness and transparency. Researchers advise for clear understanding of data used and procedure by judges to use guidance from AI systems. Explainable AI systems may help, but transparency doesn't have to mean understanding computer code.
A new study assesses ChatGPT 3.5's performance on hematology-oncology tasks, revealing its strengths and limitations in providing accurate information for patients. The study highlights the importance of physician oversight when using AI-generated medical information.
A new study from Rice University introduces a novel solar thermal-boosted organic Rankine cycle (ORC) system that can recover 60-80% more electricity annually from waste heat in data centers. The approach achieves over 8% higher ORC efficiency during sunny peak hours and lowers the cost of electricity by 5.5-16.5%.
The Wits MIND Institute has received a $1 million boost from Google.org, enabling it to drive next-generation breakthroughs in natural and artificial intelligence. The partnership aims to advance the scientific understanding of both natural and artificial intelligence, foster breakthrough research and technological innovation.
The ITU Global Symposium for Regulators 2025 endorsed guidelines to build sustainable digital ecosystems for essential national services. The new guidelines outline key tools and frameworks to drive regulatory innovation and ensure prosperity for all.
Altmetric introduces AI-powered sentiment analysis to provide deeper insights into research public response and impact. The new feature leverages advanced technology to interpret and visualize sentiment behind mentions on key social media platforms.
Researchers at Penn Engineering introduced AMP-Diffusion, a generative AI tool that designed tens of thousands of new antimicrobial peptides with bacteria-killing potential. The most potent AMPs performed as well as FDA-approved drugs without detectable adverse effects.
Researchers at the University of Minnesota developed AI-guided aerial robots to detect, track and analyze wildfire smoke plumes. The technology provides high-resolution data collection across large areas at a lower cost than satellite-based tools, enabling more accurate computer models for air quality predictions.
Florida Atlantic University's FAU Engineering has received a $827,533 USD grant from the USDA to develop an advanced edge/fog computing-based framework called FogAg. This project aims to revolutionize precision agriculture by enabling real-time sensing and analysis of crop growth and yield.
Researchers at Seoul National University of Science & Technology developed LWMalloc, a lightweight and high-performance dynamic memory allocator for resource-constrained environments. The new allocator outperforms ptmalloc by achieving up to 53% faster execution time and 23% lower memory usage.
A recent study published in Radiology found that AI-assisted mammography catches most advanced breast cancers, but still misses 14% of invasive cases. The AI missed more cases in younger women with tumors 2 cm or smaller, and those with low histologic grade, fewer lymph node metastases, and non-glandular locations.
A team of researchers at UC Santa Cruz developed a system using low-cost WiFi devices and machine learning algorithms to accurately measure heart rate. They found that the system worked regardless of body position or distance from the hardware, with high accuracy even after just five seconds of monitoring.
A new study led by Professor Jonghun Kam predicts that Pakistan will experience major floods and severe droughts on a periodic basis, exacerbated by accelerating global warming. The AI model forecasts these extreme weather events every 15 years for the upper Indus River, and roughly every 11 years for surrounding rivers.
Researchers created a new approach that integrates visual and tactile information to manipulate robotic arms, outperforming conventional vision-based methods. This breakthrough represents a significant advancement in multimodal physical AI.
Achieving universal, meaningful internet connectivity by 2030 would require a significant investment of $2.6-2.8 trillion at current prices. The report outlines the challenges, projected costs, and collaborative strategies needed to make sure everyone, everywhere, can use the Internet.
Researchers at UCLA have developed a wearable noninvasive brain-computer interface system that utilizes AI to interpret user intent, allowing participants to complete tasks significantly faster with assistance. The system demonstrates promising results for technology to assist individuals with limited physical capabilities.
Researchers have introduced a specialized AI model, EnvGPT, which achieves state-of-the-art performance in addressing complex environmental challenges. The model fine-tuned on a curated dataset outperforms larger models in accuracy and relevance, offering a scalable solution for environmental research and policy support.
Researchers at Pusan National University have developed a novel, multi-resin dispensing process for fiber-reinforced polymer fabrication, enabling precise patterning of mechanical properties within a monolithic structure. The breakthrough composite material combines flexibility and strength for advanced robotic applications.
Researchers developed DreamConnect, a brain-to-image system that integrates fMRI with advanced diffusion models. The system can interpret brain activity and edit it with natural language prompts, marking an important step in connecting human dreams with interactive visualization.
A team of computer scientists at UC Riverside has developed a method to erase private and copyrighted data from artificial intelligence models without needing access to the original training data. The approach enables AI models to 'forget' selected information while maintaining functionality with remaining data.
Researchers developed algorithms that enable robots to create the best possible outcomes from their actions while minimizing uncertainty and risk. The new strategies allow robots to anticipate human actions, adapt to different scenarios, and prioritize human safety.
A former diplomat warns that algorithms lack empathy and intuition, which are essential for successful negotiations. However, AI can streamline diplomacy and amplify human aspirations when used carefully. Diplomats need training in AI ethics and global cooperation to ensure equal access and deployment.
The procedure uses a combination of advanced imaging, artificial intelligence, and 3D printing to create a customized implant tailored to each patient's unique anatomy. This technology allows for more precise spinal alignment, reduced surgical complications, and faster recovery times.
Researchers developed AI models that can identify signs of heart failure in patients from Appalachia using low-tech electrocardiogram results. The models achieved high accuracy and could potentially provide clinicians with an edge in protecting patients' cardiac health.
Quantum Research Sciences is partnering with Purdue's Rosen Center for Advanced Computing to develop an AI-driven platform called ACID-R. The platform uses AI to quickly deliver details on commercial products and services, streamlining the process for the Air Force to identify needed technologies from the private sector.
A new AI tool has identified over 1,000 'questionable' scientific journals, including those that charge hundreds or thousands of dollars to publish research without proper vetting. The AI system uses six criteria to evaluate journals, including editorial board composition and website quality.
A new study published in Clinical Gastroenterology and Hepatology shows that AI analysis can accurately identify mucosal ulceration as well as human reviewers, providing a more objective and reproducible assessment of Crohn's disease. The findings have potential implications for education, drug development, and automated care.
Robotics experts disagree with claims that humanoid robots will surpass human capabilities in surgery, factory work, and serving as personal butlers. The main limitation is dexterity, making tasks like picking up objects difficult for current robot technology.