Articles tagged with Artificial Intelligence
An AI developed at TU Wien has shown to suggest appropriate treatment steps in cases of blood poisoning, outperforming human decisions. The AI can examine time-varying patient conditions and calculate treatment strategies, increasing cure rates by up to 3%. However, legal aspects and liability need discussion.
Researchers found that highly work-relevant personal information from an AI agent elicits greater empathy in users, while less or no disclosure leads to suppressed empathy. This study suggests self-disclosure could be used to improve people's acceptance of AI technologies.
Researchers found that machine-learning models trained with descriptive data label rule violations more harshly than humans, leading to potential serious implications in the real world. This study highlights the need for careful consideration of data labeling and training methods to ensure fairness and accuracy in AI decision-making.
The article highlights the risks of AI misuse, including its impact on democracy, job loss, and social division. Experts warn that self-improving AGI poses an existential threat to humanity, and call for regulation to minimize harm and maximize benefits.
Researchers used AI to study how the hippocampus produces varied replay types efficiently and their purpose. The model showed that sequences are prioritized stochastically according to familiarity and reward positions.
Researchers used large language models to generate open-ended answers, which were often more convincing than real responses. This approach may help gather data quickly and at low cost but also raises concerns about the authenticity of online user data.
A team of researchers developed an unsupervised entity alignment framework to improve knowledge graph search, avoiding human labor. The framework outperformed most competitors on precision and recall, scoring higher overall across multiple datasets.
A team of researchers has developed an AI tool called CRANK-MS that uses neural networks to analyze biomarkers in patients' bodily fluids and predict Parkinson's disease onset with an accuracy of up to 96%. The tool may help identify early warning signs for the disease, which can be challenging to diagnose.
Researchers used AI tools to simplify a test for hepatitis C and SARS-CoV-2, achieving 97% accuracy for SARS-CoV-2 and 95% accuracy for the most prevalent version of hepatitis C. The team plans to refine the test, improve its ability to distinguish between strains, and potentially develop at-home tests in the future.
Researchers developed EmbryoNet, an automated image analysis software that uses AI to detect and classify developmental defects in fish embryos. The software outperforms human experts in terms of speed and accuracy, making it a valuable tool for investigating the mechanisms of drug action and studying embryonic development.
Researchers from Integrated Biosciences developed an AI platform to discover novel senolytic compounds, a class of molecules targeting age-related processes. The platform identified three highly selective and potent compounds with favorable medicinal chemistry properties.
Researchers developed an online adaptive model for streaming anomaly detection that incorporates human-machine cooperation. The ISPForest method adjusts parameters and structures in real-time based on human feedback, leading to improved accuracy and reduced labor costs.
Researchers have developed a system that uses generative diffusion to create new proteins, advancing the field of generative biology. The system, called ProteinSGM, learns from image representations to generate fully new proteins, which are biophysically real and functional.
A diagnostic study using 4,095 retinal fundus images found that biomarker-based AI algorithms can be susceptible to racial bias, even when trained on raw images. This issue highlights the need for careful evaluation of AI performance in diverse populations.
The National Science Foundation has awarded Columbia University a $20 million grant to establish the AI Institute for Artificial and Natural Intelligence (ARNI), an interdisciplinary center focused on connecting AI systems to brain research. ARNI will bring together top researchers from across the U.S. to advance neuroscience, cognitiv...
The UMD-led TRAILS institute will develop AI technologies that promote trust and mitigate risks through broader participation, new technology development, and informed governance. The institute aims to create AI systems that align with values and interests of diverse groups, leading to increased transparency, reliability, and accountab...
Researchers have developed a new method called EvoAug that uses artificial DNA sequences inspired by evolution to train deep neural networks for genome analysis. This approach enables the model to recognize regulatory motifs more accurately, leading to better performance and potential breakthroughs in understanding human health.
Researchers identified three novel dual-purpose therapeutic targets using PandaOmics, which could treat both aging and glioblastoma multiforme. The target hypotheses include cyclic nucleotide gated channel subunit alpha 3 (CNGA3), glutamate dehydrogenase 1 (GLUD1) and sirtuin 1 (SIRT1).
A study published in Radiology found that AI-based decision support systems can impair radiologist accuracy on mammograms, particularly for less experienced radiologists. Even highly experienced radiologists were adversely impacted by the system's judgments, highlighting the need for safeguards to mitigate automation bias.
Researchers from Baidu Research have developed an AI algorithm, LinearDesign, that boosts COVID-19 mRNA vaccine antibody response by 128-fold. The algorithm takes a mere 11 minutes to generate the most stable mRNA sequence encoding Spike protein.
Researchers found that using positive trigger words can retrain large language models and result in less biased responses. The team analyzed GPT-2's responses to user prompts about different countries worldwide and found a significant impact on the types of adjectives used to describe citizens.
Three high school students co-authored a paper using AI engine PandaOmics to discover new therapeutic targets for glioblastoma multiforme, a common and aggressive malignant brain tumor. The study identified three genes strongly correlated with both aging and glioblastoma as potential therapeutic targets.
A study published in Radiology found that AI can analyze breast mass images from low-cost portable ultrasound machines and accurately identify cancer. The technology showed great promise for improving breast health care in low-resource settings, reducing delays in diagnosis and potentially saving lives.
Researchers at the University of Texas at Austin developed a semantic decoder that translates brain activity into text, allowing individuals with speech disabilities to communicate. The system has been trained on extensive hours of podcasts and can decode continuous language, capturing the gist of what is being said or thought.
The Connected Minds program aims to balance emerging tech's benefits and risks for humanity, with a focus on promoting equity and inclusivity. York University has received $82.8 million in federal funding through the Canada First Research Excellence Fund.
A new feature selection method for text categorization, Max-difference Maximization Criterion (MDMC), is proposed to select discriminant terms. MDMC combines a weight based on class information occupacity with ACC2 to estimate term importance, avoiding overestimation of sparse terms.
Researchers at Sainsbury Wellcome Centre found that instinctual exploratory runs enable mice to learn a map of the world efficiently. The study demonstrates how biological brains can learn faster and more efficiently than AI agents by focusing on salient objects.
A new MIT deep-learning system can analyze the internal structure and properties of materials based solely on their surface conditions. The technique uses vast amounts of simulated data to generate reliable predictions, offering a promising solution for engineers seeking non-invasive insights into material properties.
A study comparing physician and AI chatbot responses found chatbots excelled in both quality and empathy, while physicians were preferred for their nuanced understanding. The findings suggest that integrating chatbots into clinical settings could improve response quality and reduce clinician burnout.
A recent study assessed an AI chatbot's ability to answer ophthalmic questions, finding it answered approximately half correctly. The results suggest that while AI has medical advances, the current chatbot is not yet ready for substantial assistance in board certification preparation.
A team of IUPUI researchers has developed an AI-powered approach to classify insect species, tackling the challenge of discovering new species. The method uses deep hierarchical Bayesian learning to distinguish between known and unknown species, providing insight into their taxonomy and ecosystem impacts.
A deep learning AI model has been developed to screen for retinopathy of prematurity (ROP) in infants at risk of blindness. The tool was trained on images of newborns and found to be as effective as senior paediatric ophthalmologists in discriminating normal retinal images from those with ROP that could lead to blindness.
A recent study published in Sleep Medicine warns that AI-collected data may lead to misdiagnosis of childhood sleep disorders. The lack of pediatric data used to train AI models results in biased classification, causing errors in sleep stage identification and potentially severe consequences for young patients.
A study by Pohang University of Science & Technology (POSTECH) analyzed water pollution complaints in Alabama and found a significant decrease during the COVID-19 pandemic. The research team used AI methodologies to examine sentiment changes and correlations with climatic extremes.
Researchers at Argonne National Laboratory have developed a self-driving laboratory called Polybot, which automates electronic polymer research and frees scientists' time to work on tasks only humans can accomplish. The tool combines AI and robotics to streamline experimental processes and accelerate discovery.
A study found that high-quality labeling of images boosts perceptions of training data credibility, leading to increased trust in AI systems. However, biases in the data can reduce trust in certain aspects.
Researchers developed an AI-based early warning system that combines acoustic technology with artificial intelligence to classify earthquakes and determine potential tsunami risk. The system uses underwater microphones to measure acoustic radiation, which travels faster than tsunami waves and carries information about the tectonic event.
Kamaljeet Sanghera has received funding to design a summer program for high school students to learn about AI development and its ethical impact. The pilot project will utilize IBM AI kits to provide hands-on experience and technical skills training.
A massive crowd-sourced study of 327 co-authors at 186 institutions from 14 countries found that ChatGPT scored 47.4% on accounting exam questions, while students averaged 76.7%. Despite struggles with short-answer and higher-order questions, ChatGPT showed promise in improving teaching and learning processes.
A research team from Tokyo University of Agriculture and Technology has developed an image-based AI model to predict the deformation of a splashing drop. The trained encoder-decoder successfully generated image sequences that show the deformation of a drop during impact, demonstrating accurate predictions.
A €10M Horizon Europe grant will support the development of a clinical decision support tool using artificial intelligence to select the best treatment for each individual patient with high blood pressure. The project aims to improve patient care and reduce avoidable consequences such as heart attacks, strokes, and kidney disease.
Researchers have developed a computer-assisted method to automate the assessment of speech severity in ataxia patients, achieving an 80% hit rate. The new methodology leverages artificial intelligence and could simplify procedures for determining ataxia severity, facilitating research and clinical practice.
A study at ECCMID shows AI software can detect TB from chest X-rays with sensitivity and specificity comparable to that of trained radiologists. The technology has the potential to improve diagnosis in low-resource settings where radiologists are scarce.
A new machine learning framework called MILI enables accurate multi-person 3D pose and shape representation from low-resolution images. The approach tackles occlusion issues in multi-person scenes with an occlusion-aware mask prediction network.
Researchers developed an AI algorithm that allows electronic devices to express uncertainty when faced with unexpected data, improving cough detection technology. The new approach enables more precise detection with fewer sound samples per second, reducing computing power and addressing privacy concerns.
Researchers developed a nano-excitonic transistor that controls excitons to process massive amounts of data at the speed of light with minimal heat energy loss. This technology has potential applications in optical computing and realizing an era of data explosion driven by AI.
A new AI stop smoking app, Quit Sense, detected triggers and provided support to help smokers manage urges to smoke in specific locations. The study found that four times more people who used the app quit smoking six months later compared to those only offered online NHS support.
A WVU research team has created an OPTIMAL model to help students overcome fears of learning code and enhance critical thinking skills when using ChatGPT, a popular AI chatbot. The model facilitates chatbot-aided scientific data analysis and aims to improve coding skills and prompting abilities.
A team of scientists at Max Planck Institute for the Science of Light developed a method to quickly and accurately diagnose cancer using artificial intelligence and real-time deformability cytometry. The method reduces analysis time from hours to under 30 minutes, enabling faster decision-making during surgery.
Newcastle University researchers have developed environmentally-friendly photovoltaic cells that harness ambient light to power IoT devices. The cells achieve an unprecedented power conversion efficiency of 38% and are non-toxic, setting a new standard for sustainable energy sources.
The new AI system, AI-Descartes, combines theory and data to discover scientific equations, reproducing key parts of Langmuir's Nobel Prize-winning work. It also rediscovered Kepler's third law of planetary motion and approximated Einstein's relativistic time-dilation law.
Prasanna Balaprakash has been named director of ORNL's Artificial Intelligence Initiative, focusing on developing foundational, scalable, and applied technologies. The initiative aims to advance breakthroughs in materials science, biology, health science, and nuclear engineering using AI and high-performance computing.
A team of scientists created a new, fast, and precise method for analyzing cells in tissue samples from cancer patients without the need for a trained pathologist. The method uses artificial intelligence to evaluate data produced by a technique called real-time deformability cytometry.
A study published in Scientific Reports found that ChatGPT's statements can significantly influence users' moral judgements. Participants were presented with moral dilemmas and asked whether the statement they read influenced their answers.
Researchers analyzed 1.5 million patients with 53 million electronic notes to identify 9,477 distinct phenotypes and demonstrated a high degree of agreement with expert judgment in various clinical fields.
Researchers at Carnegie Mellon University developed Pix2pix3d, a machine learning tool that generates realistic 3D representations of 2D sketches. The tool allows users to create customized 3D models for various applications, including video games and consumer products.
A new study published in Nature finds that artificial intelligence (AI) surpasses human sonographers in accurately assessing cardiac function from echocardiogram studies. The research, conducted at the Smidt Heart Institute, demonstrates AI's potential to improve the quality and effectiveness of heart evaluations.
A new AI-powered worklist reprioritization tool has been shown to significantly reduce both report turnaround time and wait time for pulmonary embolus-positive CT pulmonary angiography examinations. The study found a reduction of 47.6 minutes in report turnaround time and 21.4 minutes in wait time, with no significant difference in rea...
A study by Cornell University researchers found that using AI in conversations improves communication efficiency and positive emotional language, but participants perceive their partner as less cooperative when suspected of using AI-suggested responses.
Researchers developed AI predicting heatwaves using 'deep learning' and statistical models, providing a probabilistic approach up to a month before arrival. The technology is trained on 8,000 years of weather data and combines with rare event simulation algorithms for improved forecasts