Articles tagged with Artificial Intelligence
The CONFIDENTIAL6G project aims to enhance the security and reliability of 6G networks, enabling their use in emerging AI tools and IoT devices. IMDEA Software Institute will contribute privacy-preserving computation technologies such as secure multi-party computation and homomorphic encryption.
Researchers explore how AI language models like ChatGPT understand and respond to user input, mirroring their users' intelligence. The Reverse Turing Test reveals that chatbots reflect the intelligence level of their interviewers, incorporating their biases into responses.
A novel AI architecture, relational reasoning network, accurately identifies anatomical landmarks in CT scans for orthodontic treatments. The model learns spatial relationships between landmarks without explicit image segmentation, achieving accuracy comparable to conventional methods.
The use of AI is improving diagnostic accuracy for digestive cancers, but challenges in data sharing and standardization hinder its widespread application. Researchers surveyed recent applications of AI to these deadly cancers, finding that AI can automate complex processes and detect patterns not visible to humans.
A new AI model developed at the University of British Columbia accurately predicts cancer patient survival using natural language processing to analyze oncologist notes. The model shows over 80% accuracy in predicting six-month, 36-month and 60-month survival rates.
A team of researchers at Tohoku University has developed an artificial intelligence-driven contact control system to minimize friction between moving parts in machinery, aiming to reduce wear and tear and extend lifespan.
A groundbreaking study used AI and high spatial resolution imagery to analyze over 300,000 satellite images and identify nearly 10 billion trees in sub-Saharan Africa. The research reveals that tree density and carbon stocks in the arid zone are underestimated, with a total carbon stock of 0.84 billion tonnes.
GPT-3 performs nearly on par with humans in decision-making but struggles with causal reasoning and information search. The language model's limitations may be due to its passive information-gathering approach, highlighting the need for active interaction with the world to achieve human-like intelligence.
Researchers have developed a new type of computer using human brain cells, which could lead to significant advances in computing speed and data efficiency. The technology uses small clusters of brain cells grown from stem cells and aims to increase the number of neurons to surpass electronic computers.
Researchers at Johns Hopkins University are developing 'organoid intelligence' using human brain cells, which could exponentially expand computing capabilities and solve energy consumption issues. The technology has the potential to revolutionize drug testing research for neurodevelopmental disorders and neurodegeneration.
A study by MedUni Vienna's Center for Public Health used GPT-3 to analyze the potential of AI in addressing societal megatrends, including digitization, urbanization, and climate change. The findings suggest that AI can improve understanding of these complex issues and provide suggestions for solutions.
A recent study found that Black bots were perceived as more competent and human than their White and Asian counterparts in digital interactions. This challenges past research on human-to-human interactions and highlights the importance of understanding how racial stereotypes interact with artificial intelligence.
A new cybersecurity framework uses digital twin technology, machine learning, and human expertise to detect cyberattacks in manufacturing processes. The framework analyzes continuous data streams from physical machines and their digital twins to identify irregularities and flag potential threats.
A review of 25 years of AI health tech research found that only 24% of studies include patient-reported outcomes, but there has been an increase in recent years. Experts emphasize the importance of integrating patient voices into AI models to support personalized care and prevent digitalization of healthcare.
Automated AI body composition tools achieved high technical adequacy rates (97.8-99.1%) in a large sample of external abdominal CT scans. The results support the generalizability and potential for widespread adoption of these tools, which can help build trust among radiologists and physicians.
The University of Minnesota Medical School has received $1.4 million in funding to develop methods and risk management processes for clinical artificial intelligence and machine learning to optimize patient safety. The team will focus on understanding the performance of clinical AI at the individual patient level.
A new study by NYU researchers found that infants excel at spotting motivations driving human actions, while state-of-the-art AI models struggle. By comparing infant and machine intelligence on a series of tasks, the researchers shed light on the limitations of current AI systems.
The metaverse can be used to investigate the effectiveness of large-scale interventions before implementation, saving time and money. It also provides a platform for people with limited access to natural environments to experience green spaces, reducing negative mental health effects.
A Carnegie Mellon University professor proposes a new AI model that highlights the evidence of fairness in decision-making. The model provides counterfactual explanations to demonstrate whether past decisions were fair or biased. This research has practical implications for legislative efforts, industry norms, and AI model evaluation.
Researchers used AI to analyze thousands of satellite measurements, revealing four categories of ice movements linked to meltwater flow. The study provides insight into how the Greenland ice sheet reacts to warmer temperatures and more meltwater.
The Heidelberg Laureate Forum Foundation offers journalist travel grants to cover the 10th HLF from September 24-29, 2023. Recipients of prestigious awards in mathematics and computer science will gather with young researchers for a week of interdisciplinary dialogue.
A new study by Tulane University demonstrates that even a single atom can act as a reservoir for computing, processing information optically. The researchers proposed a non-linear single-atom computer where input and output are encoded in light, enabling flexible computation with any desired outcome.
Researchers develop a governance model for ethical guidance in AI, combining copyleft licensing with the patent troll approach. The CAITE model assigns enforcement rights to a third-party host, ensuring compliance while promoting flexibility and community participation.
Researchers at University of Gothenburg developed AI method using graph theory and neural networks to analyze cell movement, enabling better understanding of biological processes and development of new medical technologies. The method can reconstruct cell paths and test medication effectiveness as potential cancer treatments.
A Penn State-led research team found that language models can plagiarize content in three ways: verbatim, paraphrase, and idea reuse. The study highlights the need for more research into text generators and their potential ethical implications.
Scientists have developed a simulation environment to test multistage attack scenarios using deep reinforcement learning, achieving success rates of up to 95% in stopping cyberattacks. The AI-powered approach offers promise for proactive cyber defense and the ability to detect changes in the cyber landscape earlier.
A recent study published in Human Reproduction found a significant association between prenatal alcohol exposure and changes in children's facial features. The research used AI to analyze three-dimensional images of children's faces at ages nine and thirteen, revealing that mothers' alcohol consumption before and during pregnancy can h...
Researchers at the University of Konstanz found that selfish behavior can lead to a fair distribution of risk within groups of animals. By positioning themselves optimally to minimize distance from others, individuals reduce their own predation risk and create a safer environment for all members.
Researchers at Cornell University developed an AI tool to track online debates and detect when tensions are escalating. ConvoWizard provides real-time feedback to users on potentially incendiary language, encouraging constructive debate.
A University of Kansas study analyzed over 1,000 AI-generated ads and found that they are only labeled as ads about half the time, often appealing to consumers in positive ways. The prevalence of AI in programmatic advertising raises concerns about consumer deception and the need for updated guidelines to require transparency.
Researchers use movie clips to study mouse brain activity, revealing novel ways to diagnose learning and memory disorders. The discovery points to new ways to tackle Alzheimer's and improve artificial intelligence.
A new clinical trial will assess the efficacy of an AI-powered massage robot, EMMA, in managing chronic low back pain. The study aims to explore the effectiveness of robotic Tuina therapy in alleviating pain and improving health outcomes.
Researchers at MIT developed a technique to improve machine-learning models' reliability without requiring additional data or extensive computing resources. The method uses a simpler companion model to estimate uncertainty, enabling more effective uncertainty quantification.
A new study uses Fourier analysis to understand how deep neural networks learn complex physics. By analyzing the equation of a fully trained model, researchers were able to identify crucial information about how the network learns and generalizes. This breakthrough could accelerate the use of scientific deep learning in climate science.
Research from the University of Cambridge found that cinematic depictions of AI scientists are heavily skewed towards men, contributing to a cultural stereotype that may deter women from pursuing AI careers. The study highlights the need for more diverse representation in AI development.
Scientists developed an automated algorithm using machine learning to identify image content related to illegal wildlife trade in digital spaces. The method uses feature visualization and achieves high accuracy in distinguishing between natural and captive contexts.
A study published in PLOS Digital Health found that ChatGPT scored between 52.4% and 75.0% on the USMLE exams, exceeding the passing threshold of approximately 60%. The model demonstrated high concordance rates and produced significant insights for most responses.
Researchers at Brigham Young University have developed an AI algorithm that can accurately locate players and determine formations in football game footage. The system has achieved over 90% accuracy on player detection and 85% accuracy on formation identification, potentially eliminating the need for manual annotation and analysis.
The study uses artificial intelligence to generate new molecules with optimized properties, applicable to various industries. The method enables finding Pareto-optimal solutions, leading to more efficient materials for optoelectronics, solar energy, and other fields.
Researchers tested three common techniques to make algorithms fairer and found that one approach didn't reduce social norm bias at all. They proposed a new technique: a formula to directly measure social norm bias in an algorithm so it can be corrected. This bias can persist even after overt discrimination is removed.
A team of researchers developed a model-free approach using deep reinforcement learning to optimize estimation of multiple parameters in quantum sensors. The protocol achieved significantly better estimations compared to nonadaptive strategies, demonstrating enhanced performance in resource-limited regimes.
Researchers found pigeons can solve complex categorization tests through exhaustive trial and error, mirroring artificial intelligence's process. Pigeons' ability to memorize patterns and objects surpasses human performance in similar tasks.
FRIDA uses AI models similar to those powering tools like ChatGPT and DALL-E 2 to generate paintings based on user input. The robot's final products are impressionistic and whimsical, with bold brushstrokes that lack precision sought in robotic endeavors.
Engineers at the University of Waterloo developed AI technology to predict breast cancer treatment effectiveness, allowing doctors to prescribe personalized treatments and reduce unnecessary side effects. The new algorithm uses synthetic correlated diffusion imaging images to forecast patient responses to chemotherapy.
The Human Brain Project Summit 2023 will highlight the scientific achievements of the project at the interface of neuroscience and technology. Researchers will present cutting-edge research on European advances in neuroscience and applications in medicine and brain-derived technologies.
Researchers developed an AI algorithm to predict prime editing efficiency and accuracy, enabling faster and more precise genome editing. By analyzing a large data set of over 100,000 pegRNAs, the tool identifies key properties influencing prime editing success.
A new study reveals that online hate speech is frequently conveyed through subtle and modified wording, allowing it to evade detection. Covert methods of expression are being used by haters to avoid detection, highlighting the need for increased attention to implied ways of expressing hate.
Researchers developed a machine learning model using advanced 2D chemical descriptors to predict highly selective asymmetric catalysts without quantum chemical computations. The model demonstrated high accuracy in predicting catalyst structures and selectivity, outperforming existing methods.
A team of researchers has proposed a novel method for multi-label classification that enables continual learning from big data. The approach combines clustering and Bayesian methods to achieve high classification performance and adaptability in the face of new data.
A new approach to deep reinforcement learning demonstrates ability to stabilize large datasets used in AI models, which may lead to uncovering ways to arrest cancer development. The method has been successful in designing and refining existing therapies, with the next step being to use live cells.
Researchers have developed a compact silicon photonic compute engine capable of computing tiled matrix multiplications at a record-high 50 GHz clock frequency. This achievement promises to contribute significantly to data center cybersecurity and enables real-time threat detection for malicious packets.
The University of Alberta and Amii are recruiting 20 new Canada CIFAR AI Chairs to focus on groundbreaking AI research in health, energy, and Indigenous initiatives. This initiative aims to accelerate opportunities for responsible commercialization of AI, driving future economic growth.
A new study at Cornell University introduced an automated participant that periodically interrupts the conversation to give nonnative speakers a chance to speak. The AI bot increased participation from 12% to 17% of all words spoken, with nonnative speakers feeling valued and appreciated for their perspectives.
A recent study found that AI technologies exhibit more pronounced biases in facial age perception than humans, with exaggeration of smiling faces' aging effect and decreased accuracy for older adults. The research suggests these biases stem from visual cues rather than preconceptions.
The JIPipe software enables automated analysis of images generated in research without requiring programming skills. Users can create flowcharts and perform automatic image analyses using artificial intelligence.
The use of artificial intelligence (AI) and language models in writing manuscripts has sparked debates about the integrity of scientific publication. AI-generated content can lead to misinformation and compromised academic standards, posing significant risks to medical knowledge.
Researchers have developed an AI-based resource to assist individuals in identifying recommended actions based on their clinical profile and COVID at-home test results. The system uses a combination of symptoms and home tests to provide more accurate diagnoses and improve patient care.
UC San Diego scientists developed an AI-based strategy for discovering high-affinity antibody drugs, identifying a new antibody that binds 17 times tighter than an existing one. The approach accelerates the discovery process, reducing failure rates and enabling faster development of novel cancer and COVID-19 treatments.
A deep learning model distinguishes colon carcinoma from acute diverticulitis in CT images, improving diagnostic performance and reducing false-negative and false-positive findings. The model's sensitivity and specificity were significantly improved, supporting radiologists in patient care.
Researchers identified a panel of novel blood biomarkers associated with sudden cardiac arrest, which have the potential to enhance clinical prediction. Additionally, an AI-powered ultrasound technology can accurately identify coronary artery calcium buildup, potentially replacing CT scans as a diagnostic tool.