Articles tagged with Artificial Intelligence
The study reviews SLM optimization strategies and evaluates a Greek labor-law assistant that combines fine-tuning, Retrieval-Augmented Generation and quantization for practical local deployment. The evaluation used standard language-generation and retrieval-oriented metrics, indicating that larger models achieved stronger scores while ...
Researchers emphasize the importance of enhancing visual intelligence in autonomous robots to achieve complex tasks. Visual perception, decision-making, path planning, and control must be integrated for robots to perform reliably. The ultimate goal is to make robots have human-like intelligence and enable collaboration between robots.
A new study reveals that large language models can systematically alter the direction of users' messages on contested topics, even when instructed to preserve the original meaning. The researchers show how these small changes can accumulate and gradually influence broader public opinion over time.
Researchers developed an on-chip all-optical supernode for ultra-low-latency deep neural network inference, achieving a 100-fold increase in inference speed while using only one-ninth of computing resources. The system supports high-speed data routing and switching with low loss and flat response over a spectral range exceeding 100 nm.
Researchers propose a new framework, BOC, inspired by biological nervous systems to build more adaptive intelligent systems. The framework integrates sensing, memory, and computation directly at the hardware level, reducing data movement and latency.
Prof. Orly Lewis is developing a flexible publishing platform for interactive knowledge environments, while Prof. Nir Friedman is creating an epigenomic liquid biopsy for early detection and monitoring of Metabolic Dysfunction-Associated Steatohepatitis.
A new AI model, SpliceSelectNet, accurately predicts RNA splicing by capturing long-range DNA signals. The model's hierarchical Transformer architecture preserves high computational efficiency while maintaining single-nucleotide resolution, enabling accurate analysis of genomic regions.
A new method uses machine-learning algorithms to detect scatterable PFM-1 landmines over wide areas, providing a first-pass analysis of suspected hazardous areas. The technique requires only a lightweight laptop, drone, and camera, operating without an internet connection.
A chatbot using GPT-4o technology delivers personalized messages to American adults, increasing support for transgender rights and improving attitudes, but effects are short-term. The intervention demonstrates AI's potential as a scalable complement to community-based approaches aimed at fostering inclusion.
Researchers at ISTA have developed a way to guide AlphaFold with experimental data, enabling the model to better reflect physical and biological reality. This approach aims to improve future predictive models by accounting for structural heterogeneity and dynamism in proteins.
Physicists have identified two new superconductors, YRu3B2 and LuRu3B2, using machine-learning to filter material combinations. This breakthrough aims to find a room-temperature superconductor, which could slash global energy consumption and reduce the heat footprint of ICT sector.
University of Jaume I students Pau Montagut and Mario García won first place at the ICRA 2026 robotics conference with an AI model that taught a Toyota HSR robot to perform household tasks. The team's achievement is notable given their undergraduate status, competing against teams of research personnel.
The new photonic architecture harnesses three fundamental degrees of freedom: wavelength, mode, and polarization, achieving 192 parallel computing channels. The chip supports large, reconfigurable convolution kernels up to 13x13, capturing global structural contours while preserving fine details.
Researchers warn that teens' reliance on AI chatbots may bypass opportunities for developing essential relationship skills. The technology offers immediate, nonjudgmental guidance but lacks needed safeguards, potentially reinforcing unhealthy relationship patterns and increasing vulnerability to mental health problems.
The Hebrew University of Jerusalem and Edut 710 have partnered to create a groundbreaking AI-powered living archive, featuring nearly 2,000 survivor, witness, and first responder testimonies. The archive will enable users to search, translate, and explore the accounts through natural language, preserving authenticity and integrity.
A large clinical trial found that an AI support tool improved the quality of clinical documentation and treatment planning, but did not significantly change short-term patient outcomes. The study involved over 9,600 patients in Kenya and used a randomized controlled design to test the effectiveness of the AI tool.
Stanford researcher Ellen Kuhl's AI tool, BurgerAI, creates novel burger recipes optimized for deliciousness, sustainability and nutrition. The burgers were tested in a blinded taste test and found to be comparable to popular fast-food options.
Digital twins are expanding rapidly, using real-time data to simulate and analyze systems before applying them in the real world. The study emphasizes the need for interoperable architectures, machine-readable metadata, and standardized trust frameworks to address challenges such as privacy, cybersecurity, and uncertainty.
The rise of AI-generated content threatens authenticity in the clinical landscape, from AI chatbot impersonations to deepfakes of real physicians. Existing laws and regulations are struggling to keep pace with these emerging challenges.
The review discusses key AI concepts, including algorithms, models, architectures, machine learning, deep learning, and multimodal models. It highlights their clinical applications, such as detection of lymph node metastases, Nottingham grading, biomarker quantification, risk stratification, and prognostic prediction.
The University of Phoenix College of Doctoral Studies has published a new white paper examining the growing AI fluency gap in the workplace. The paper argues that AI fluency is no longer only a productivity issue but a retention issue, as employees are rapidly building AI skills while organizations struggle to keep pace.
Artificial intelligence is being harnessed to integrate multi-omics data from large-scale repositories such as TCGA and GDSC to decode tumour drug resistance. The review advocates for explainable AI frameworks, multimodal fusion strategies, and dynamic liquid-biopsy monitoring to address barriers to clinical adoption.
Researchers developed an AI framework to identify viable target antigens for CAR T cell therapy, using publicly available single-cell RNA sequencing data. The framework showed robust tumor-killing activity in mouse models of multiple cancer types, including melanoma, leukemia, and colorectal cancer.
Researchers developed a new system called Murakkab to optimize agentic workflows for AI applications. It enables developers to describe their intent in high-level terms, automating the selection of models and tools to use, and optimizing hardware configurations in real-time.
The study successfully integrates probabilistic sampling and deterministic computation in generative AI hardware within a single ferroelectric memory array. The technology enables the generation of diverse images reflecting facial attributes, improving area and power efficiency in applications.
The EU project BRISA aims to test safe and clean industrial technologies that reduce risks to human health and biodiversity. The project will protect over 45,000 workers by 2035 and prevent hundreds of premature deaths each year.
Researchers at the University of Washington created PaperTok, a tool that uses AI to help scientists and researchers create short-form videos from research papers. The system generates scripts and video clips in 45 seconds, making it easier for non-experts to engage with scientific content.
JMIR Publications and Jisc have expanded their Flat-Fee Unlimited Open Access Partnership, making APCs free for researchers affiliated with participating institutions. The agreement allows researchers to prioritize open access publishing without financial burdens.
Researchers developed a tandem neural network that rapidly infers key semiconductor material properties from simple transistor measurements, outperforming conventional approaches. The system produces results in under one millisecond with near-perfect accuracy.
A new study has unveiled a high-resolution global map of seagrass, revealing areas of loss and recovery, and providing valuable insights for conservation efforts. The map shows that seagrass ecosystems store approximately 640 trillion grams of carbon, equivalent to the yearly emissions of 500 million cars.
A new article by Trevor van Mierlo highlights the importance of trust in digital behavioral health. Key findings include the need for dynamic governance and verifiable audit trails to address mistrust issues.
Researchers Anna Galler and Bettina Könighofer at Graz University of Technology will use FWF's Astra awards to identify quantum materials for future electronics and develop trustworthy AI systems. Their projects focus on protecting AI systems from risky actions and exploring new materials with unique electronic properties.
A new commission, launched at the Texas Brain Economy Summit, aims to define, measure, and operationalize brain capital, a concept encompassing both brain health and critical cognitive skills. The initiative seeks to address the gap in economic planning and promote brain health as a driver of economic resilience.
Researchers at Tohoku University have created a clearer map for searching for hydrogen storage materials, identifying key physical factors that control their performance. The study suggests adjusting geometry and lattice flexibility to raise capacity while tuning stiffness to keep equilibrium pressure near everyday conditions.
Researchers used AI to analyze mammograms and found that women who developed breast cancer had increasing risk scores over time, while those who did not had stable scores. The study suggests that image-based AI risk scores can predict future breast cancer risk in women without a known genetic mutation or family history.
A self-driving chemistry lab called Flex-Cat has been developed to autonomously search for faster and more selective ways to make important industrial chemicals. The platform combines robotics, high-pressure reactors, and artificial intelligence to identify high-performing catalysts and those that can be programmed to produce different...
Researchers developed RNovA algorithm to identify new PTMs in human cells, expanding capabilities of machine learning in basic biological research. The discovery aims to advance diagnostics and broaden biologists' horizons for cancer and other diseases.
Researchers developed an algorithm, RAPID, to re-identify wild animals using their coat patterns. The algorithm achieved high accuracy rates and demonstrated its speed on various datasets, making it a promising module for wildlife monitoring and ecological analyses.
Researchers have developed a versatile uncertainty-aware AI framework called TRUECAM that provides customizable accuracy guarantees for cancer subtype classifications. TRUECAM outperforms existing approaches to digital pathology AI uncertainty quantification, detecting out-of-scope inputs and improving fairness across sex and race.
A University of Houston engineering professor developed a mathematical model to help decision-makers decide where to spend limited dollars on infrastructure resilience. The model accounts for real-world uncertainty and identifies critical assets to invest in, providing the greatest benefit before disaster strikes.
Thomas Pock's ERC Advanced Grant aims to develop novel generative learning methods and algorithms for computer vision, improving medical imaging by understanding and realistically generating images. The project seeks to establish a close link between data analysis and generation.
A study of funded AI startups reveals occupations with high AI exposure, including office clerks and data scientists, while manual tasks like construction work are less affected. The Occupational AI Startup Exposure (AISE) index also notes that jobs requiring social skills or ethical decision-making may be less likely to automate.
Researchers developed a method combining AI and mathematical models to accelerate MRI scans for breast cancer imaging, achieving improved tumor visibility and high diagnostic sensitivity. The ELITE method has the potential to improve not only MRI scans but also other imaging platforms.
MIT researchers developed a new system-on-a-chip called Gleanmer, which generates highly accurate 3D maps of the robot's environment using Gaussians to represent obstacles. This approach reduces memory and power consumption by up to 99%, making it suitable for lightweight augmented reality headsets.
The US retains its scientific edge but is losing the race to translate discoveries into cures, warns a new report from Cure Innovation Index. Without immediate renewed investment and policy changes, the US scientific edge will not hold.
Researchers develop AI framework using principles of superposition and entanglement to tailor cancer treatment to patients' entire molecular background. The technique predicts health outcomes and suggests genes to target, outperforming standard biomarkers in clinical trials.
The report reveals cutting-edge technologies affecting physical systems, from energy to medicine and manufacturing. These innovations have the potential to make a meaningful impact on pressing challenges like climate change and food insecurity.
Primary and secondary schools in Shanghai are using an AI-based classroom analysis system to support instructional improvement. The platform automatically processes video data and provides theory-informed analytic reports, enabling teachers to track student agency and professional growth over time.
A landmark international competition has revealed that AI systems designed to spot fake faces perform unevenly across demographic groups, with lighter-skinned individuals enjoying higher accuracy while darker-skinned faces are more frequently misclassified. The top-ranked solution combined data curation, mixture-of-experts architecture...
The new report identifies six principles shaping trust in AI: reliability and competence, contextual awareness, transparency and accountability, fairness and integrity, resilience, and relational dynamics. The paper argues that trust must be grounded in demonstrated system performance, governance, and institutional responsibility.
Researchers developed a data-driven method combining GA-BP neural network and chaotic particle swarm optimization to predict and optimize screen-printing parameters for thick-film resistors. The approach achieved an R² of 0.991, recommending optimal settings in 2.38 seconds and reducing resistance deviation within 5%.
A groundbreaking technology called Time to Move (TTM) offers unprecedented control over object and character movement in AI-generated videos. TTM eliminates the need for complex infrastructure or training on millions of videos, making AI video creation more accessible.
A new framework proposes a third way to balance AI innovation and safety: accelerating responsible innovation through technical, organizational, and ethical advancements. The Telus GenAI customer support agent demonstrates how risks can drive further innovation, reducing the need for restrictive safety constraints.
A new research direction proposes building machine-learning systems on top of AI models to detect hidden information and predict behavior. This enables users to supervise the model, control its behavior without understanding the entire mechanism.
Researchers create PhishLumos AI system to detect phishing campaigns by analyzing infrastructure clues, achieving 8-day faster detection than experts. The system uncovered over 190,000 new links, with 92% later flagged as malicious, outperforming content-centric approaches.
A new AI model analyzed 30 years of EHR data to identify patients with primary aldosteronism, a common cause of high blood pressure. The model correctly flagged over 90% of cases while missing fewer than 10%, suggesting it could improve screening for this condition.
The system uses two AI agents to comb through publicly available data and conduct life cycle assessments, achieving an average error rate of 5%-19% similar to expert-led LCAs. The team also developed a new method to bypass detailed data collection and estimate carbon footprints for unknown devices using 'nearest-neighbors' approach.
The study synthesizes recent advances in single-cell and spatial transcriptomics to identify tumor-enriched cell subsets closely related to prognosis and treatment response. The review introduces the
Rockefeller University Press has partnered with Cashmere to enable safe and transparent AI inference use of its journals, protecting authors' work while providing visibility and revenue transparency. The partnership uses Cashmere's infrastructure suite to manage AI-powered research applications.
Researchers at Princeton University used AI to analyze how drugs affect cell structures, finding new shapes linked to disease and discovering a novel drug effect. The neural network identified cap, necklace, and flower shapes, with the latter indicating a previously unknown role of an enzyme in maintaining nucleolar organization.