Articles tagged with Artificial Intelligence
Experts say AI's portrayal as White removes people of color from the future, perpetuating racial biases in algorithms and workforce. Cultural depictions of AI as White risk creating a 'racially homogenous' technologist workforce.
A new viewpoint piece in JAMA outlines key considerations for implementing chatbots in healthcare, including patient safety, trust, and transparency. The authors propose a framework to inform decision-making on chatbot implementation, emphasizing the need for robust evaluation and research.
A group of doctors and data scientists propose creating clinical departments devoted to artificial intelligence to harness its power in transforming patient care. The plan aims to address issues such as inconsistent implementation, poor data quality, and overdependence on AI, and would bring together diverse expertise for faster benefits.
A multidisciplinary group of Indigenous scholars, including professors and graduate students from Canada and the US, met in Hawai'i to brainstorm how Indigeneity intersects with AI. Their completed position paper offers diverse perspectives on what AI means and offers to Indigenous peoples, challenging long-held assumptions about Weste...
QUT researchers have published a new method for detecting koalas in bushfire areas, using drones equipped with infrared cameras and an artificial intelligence algorithm. The system is more accurate and less invasive than traditional methods, providing a more reliable estimate of the population size.
A team of researchers developed an AI model to forecast complicated large-scale tropical instability waves in the Pacific Ocean. The model, which uses satellite data-driven deep learning, accurately predicted sea surface temperature evolution and captured TIW propagation's spatial and temporal variation.
Researchers used drones with thermal cameras to locate ground-nesting farmland bird nests, identifying a promising conservation tool. The system performed best under cloudy and cold conditions, showing potential for automated nest detection in agricultural practices.
UT Dallas researchers are using the Minecraft mod Polycraft World to simulate dynamic and unexpected events to train AI systems. They aim to create artificial intelligence agents that can respond appropriately in novel scenarios, such as changing weather or unfamiliar terrain.
A new German research project aims to improve the automatic analysis and generation of high-quality texts using artificial intelligence. The goal is to disentangle styles and topics in textual data, enabling enhanced text quality and influencing tone, while overcoming challenges with irony and sarcasm.
Researchers have found a mathematical means to help regulators and businesses manage AI systems' biases towards making unethical and costly commercial choices. The new 'Unethical Optimization Principle' provides a simple formula to estimate its impact, aiming to reduce the risk of AI-driven unethical strategies.
AI systems can learn to recognize abnormalities, but often lack transparency and make 'lazy' diagnoses. Researchers from Radboud University Medical Center developed an approach to improve sensitivity and transparency in AI diagnosis, achieving a 11.2% increase in diabetic retinopathy detection
Researchers at University of Cambridge developed a context-aware AI system that reduces the 'communication gap' by up to 96% for nonverbal people with motor disabilities, enabling faster conversation rates.
A new AI-driven system called ModGraProDep has been developed to predict cancer survival and risk. This innovative tool uses probabilistic modeling and machine learning to provide personalized predictions for patients, enabling more effective treatment planning and improved patient outcomes.
A team of scientists used artificial intelligence to examine whether mandrills' facial features are a result of genetic similarity or selection. They found that half-sisters with the same father exhibit stronger facial resemblance, indicating a possible evolutionary mechanism for social discrimination and nepotism among kin.
The Lancet review calls for guidelines and research agenda to ensure equitable and ethical use of AI in low- and middle-income countries. The recommendations focus on human-centered design, data protection, and regulatory standards.
Researchers developed an intent-defined optical network (IDON) with artificial intelligence-based automated operation and maintenance to address the challenges of manual configuration. IDON achieves intent-oriented configuration conversion, adaptive generation and optimization strategies, and closed-loop intent-guaranteed operations, d...
A scoping review of 405 studies reveals that AI research in primary care is at an early stage, with unique challenges to applying AI in a person-centric approach. To mature, value must be placed on developing rigorous AI and identifying impacts on care delivery and health outcomes.
A recent study found that artificial intelligence can provide results comparable to lung function tests in diagnosing emphysema and classifying its severity. The AI system can automatically analyze chest scans and create a 3D model of the patient's lungs, helping doctors better communicate with patients and track treatment progress.
Researchers found that AI systems can absorb moral responsibility in difficult conversations, allowing humans to trust them more than their human partners. The study explores how AI alters human interactions and suggests potential applications for improving communication with subtle guidance from AI.
Researchers warn of 'exaggerated' AI study claims, highlighting poor quality evidence and methodological flaws in studies. The majority of deep learning medical imaging studies have high bias and lack transparency, fueling hype for rapid implementation.
CURATE.AI is an AI-derived mechanism-independent technology platform that profiles individual patients based on their data and recommends optimal drug doses. This approach aims to improve treatment efficacy while reducing toxicity levels and addressing the challenges of personalized dosing.
Scientists at Osaka University created a flexible computing device from field-programmable gate arrays, increasing circuit density by 12 times and reducing energy usage by 80%. This innovation may lead to custom machine learning hardware with enhanced performance and reduced electricity consumption.
Researchers at Aarhus University have successfully automated the process of measuring and documenting terrain using drones, making it faster, cheaper, and easier. The AI system takes control of the drone's movement, adjusting its distance from the wall and camera position to capture high-quality images with minimal human intervention.
Researchers at Mount Sinai have developed an AI model that can accurately determine the type of lower back pain by analyzing clinical notes in electronic medical records. This breakthrough could lead to improved treatment options, reduced opioid prescriptions, and more accurate billing and therapy recommendations.
Researchers developed an automated AI technique to analyse CMR scan images, precisely quantifying blood flow to the heart muscle. The results showed a strong correlation between reduced blood flow and adverse health outcomes, including death, heart attack, stroke, and heart failure.
The selected scientists and engineers will participate in a week of intensive public engagement and science communication training, networking, and plan development. The fellows will return to their institutions with resources and connections to develop, implement, and evaluate their public engagement activities.
A 100-plus page report presents a vision for the strategic path to unleashing AI's full potential, with applications in healthcare, education, business innovation, and more. The roadmap aims to address the tremendous social change that will result from AI research.
A Tufts University team won a grand prize in the NSF Idea Machine Competition with their project 'From Thinking to Inventing,' focusing on what machines can invent and how. They aim to create resilient robots that can creatively solve problems, addressing pressing issues like climate change and wealth inequality.
Researchers at KAIST discovered a computational and neural mechanism for human meta reinforcement learning, enabling the porting of key human intelligence elements into AI algorithms. The study highlights how humans adapt to complexity and uncertainty in decision-making.
A new study uses an interactive voice application with AI to monitor patients' mental health, offering personalized analysis and finding accuracy comparable to physician assessments. The application, MyCoachConnect, analyzed patients' speech patterns and found unique changes over time, helping track mental states.
Researchers at the University of Warwick developed an AI technology that can detect hypoglycaemic events from raw ECG signals using wearable sensors with 82% reliability. This non-invasive method could replace fingerprick testing, particularly useful for pediatric patients.
Researchers at Universidad Complutense de Madrid developed an AI-powered algorithm that uses deep learning to detect irregularities in rice labelling. The algorithm achieved precision models between 93 and 99%, showing potential for future applications in the food industry.
The MIPT team developed a new method for reinforcement learning with demonstrations, enabling rapid solution of hierarchical problems in Minecraft. This approach opens opportunities for applying reinforcement learning to real-world tasks like autonomous vehicle control and manipulator control.
The Journal Healthcare Transformation: Artificial Intelligence, Automation, and Robotics explores the challenges and opportunities for integrating technology into consumer healthcare. A recent article found that five telemedicine companies provided nearly half of telemental health visits, highlighting the need for coordinated care.
The mind composes meanings from separate words through a dynamic interaction between brain regions, using multi-modal cues. Humans and artificial intelligence systems differ in representing meaning, with humans using contextual knowledge and emotional cues.
Researchers at the University of Waterloo developed an AI tool that uses deep-learning algorithms to determine if claims are supported by other posts and stories on the same subject. The system achieved 90% accuracy in stance detection, a key area of research in detecting fake news.
Researchers have developed an algorithm capable of identifying who in a social group would be the best persons to be trained as 'gatekeepers' capable of recognizing warning signs of suicide. The tool aims to improve efficiency of suicide prevention trainings for vulnerable populations, such as homeless youth.
Researchers at Stevens Institute of Technology have completed the first census of Wikipedia's 1,601 bots, shedding light on their functions and interactions with human users. The study found that bots play nine core roles, including fixers, connectors, and protectors, which together account for about 10% of all activity on the site.
The MIT researchers developed a bot equipped with artificial intelligence that can beat human players in tricky online multiplayer games where player roles and motives are kept secret. The bot uses deductive reasoning to identify friend or foe, ensuring victory over humans in certain games.
Researchers develop AI-driven platform to design, build, test and learn complex biochemical pathways for efficient production of lycopene, a red pigment used as food coloring. The system reduces time and cost by narrowing down possible pathways from over 10,000 to 100.
A recent study by NYU Abu Dhabi researchers found that bots posing as humans are more efficient at persuading partners to cooperate in a game. However, once their true nature is revealed, cooperation rates drop and their superiority is negated.
Researchers at Thayer School of Engineering at Dartmouth have developed an approach for detecting a speaker's intent to mislead, which could be used to extract opinion from 'fake news.' The framework uses universal features of deceptive reasoning and has been shown to outperform models based on verbal changes and personal differences.
A University of Texas at Austin team developed an AI system named Fashion++ that analyzes outfits and suggests tweaks for a more fashionable look. The tool uses visual recognition systems to consider color, pattern, texture, and shape, offering alternative outfits to users.
A new study published in INFORMS Journal Marketing Science found that AI chatbots can increase sales by four times compared to inexperienced workers. However, customer awareness of the use of AI chatbots before making a purchase leads to a significant decline in sales rates.
Samantha Kleinberg is developing artificial intelligence that provides personalized information to patients, enabling them to actively manage their health. The research focuses on using wearable data and machine learning to identify causes of diseases and develop more usable algorithms that consider human factors.
A multidisciplinary team from NYU Tandon School of Engineering is conducting foundational research to understand the impact of AI on expert work in healthcare. The project aims to redesign technology, workflow, and interactions to improve the effective use of healthcare technologies and human capabilities.
Researchers used a newly developed algorithm to study mixed x-ray images of the Ghent Altarpiece, separating features from the front and back of the painting's double-sided panels. The analysis improved our understanding of art masterpieces and provided new opportunities for art investigation, conservation, and presentation.
A new study aims to explore ways to share decision-making and information between humans and AI drivers, addressing concerns about perceived loss of control and fear of unknown actions. The goal is to increase user trust, consumer acceptance, and overall usability of autonomous vehicles.
An AI tool developed by USC computer science student Timothy Greer classifies song genres by analyzing lyrics and chords. The tool predicts 'Old Town Road' is country according to lyrics, rock according to chords, and pop when combining both. This method highlights the complexity of music perception and its relation to music construction.
A new study uses AI to predict food product recalls from Amazon reviews with 74-percent accuracy, identifying thousands of potentially unsafe products. The researchers trained an existing AI called BERT on customer reviews to identify recalled foods and found terms associated with FDA recalls in 20,000 other reviews.
Experts Eric Topol and Kai-Fu Li argue that international collaboration outweighs confrontation in the development of artificial intelligence for healthcare. They critique the forced divestiture of iCarbonX as misguided and highlight the need for access to clinical data to compete in AI health.
Researchers developed Pluribus AI that can play six-player no-limit Texas hold'em poker, outperforming top professionals and five copies of itself. The AI confirms traditional wisdom on limping as a suboptimal strategy for most players.
The National Science Foundation is funding a new supercomputer at the Pittsburgh Supercomputing Center (PSC) with a $10-million grant. The Bridges-2 system will provide massive computational capacity and flexibility to researchers in Pennsylvania and nationwide.
Scientists translate amino acid sequences into musical compositions, then use AI to generate new proteins with desired features. The method also enables the creation of bio-based musical pieces.
Recent Chinese AI research has made significant breakthroughs in big data analysis, with a focus on developing algorithms to detect abnormal data regions and repair historical correlations. Researchers have also proposed using Doppler measurements to improve target tracking performance in noisy environments.
Researchers developed an AI method to recognize patterns in infant cries, enabling healthcare workers and parents to accurately determine a baby's needs. The algorithm uses compressed sensing to process noisy environments, allowing for independent use across various scenarios.
The Radio Me project uses AI to adapt and personalize live radio, providing personalized reminders, information, and music for people with dementia. The system aims to improve quality of life and reduce hospital admissions by up to 30%.
Researchers found that when participants were unsure whether a profile was written by a human or an algorithm, they distrusted the one they believed to be machine-generated. The study suggests designing AI communication tools with transparency and trustworthiness in mind can improve user experience.
The Argonne National Laboratory's Aurora supercomputer will revolutionize scientific research and discovery with its exaFLOP performance and ability to handle both HPC and AI. The system is expected to have a significant impact on various fields, including cancer research, climate modeling, and veterans' health treatments.
The lack of regulatory standards in clinical artificial intelligence poses concerns about safety and efficacy, according to a recent study. The authors propose five standards to guide evaluation and regulation of clinical predictive analytics, emphasizing the importance of established clinical standards and clinician agency.