Articles tagged with Artificial Intelligence
The Tong test proposes five critical characteristics for AGI benchmarks: infinite tasks, self-driven task generation, value alignment, causal understanding, and embodiment. This innovative approach combines a general algorithmic testing paradigm with human-AI interaction-based testing.
Scientists developed computational eye models to help patients and surgeons select ideal intraocular lenses and predict visual outcomes. The technology uses anatomical information of the patient's eye to provide guidance on expected optical quality post-operatively.
A comprehensive survey of ChatGPT papers reveals a rapid increase in interest for NLP models, but also raises concerns about biased content generation and technology misuse. The study emphasizes the need to establish responsible development and deployment frameworks for ChatGPT.
A new deep-learning model was developed by Professor Sanghyun Park's research team at DGIST to classify brain waves with high accuracy using only a small amount of data. This model is capable of overcoming the limitations of existing classification models and can be used for future research related to brain waves.
Researchers at U of A are developing an AI-powered system to analyze cardiac images, enabling clinicians to diagnose heart disease more accurately. The project aims to reduce healthcare costs and improve patient outcomes, with potential global implications.
Researchers used AI to predict 3D shapes of over 215 million proteins, providing insights into their functions and evolution. The Protein Universe Atlas offers a valuable resource for scientists to explore protein diversity.
A study funded by a $180,000 grant from the National Science Foundation will investigate sleep and sleep disorders using RFID tags attached to clothing. The research aims to detect abnormal body movements that may indicate health problems, such as falls in elderly individuals.
The Open Encyclopedia of Cognitive Science (OECS) is a dynamic web reference that will equip readers with essential tools to grapple with the profound implications of cognition and intelligence in today's society. With generous funding from James S. McDonnell Foundation and the Allen Institute for AI, the first set of articles will be ...
The project aims to assess the operational resilience of microgrids on DoD installations and ships, using new operational resilience indexes developed by Lehigh University researcher Javad Khazaei. The team will develop a dashboard to monitor resilience indexes in real-time, providing recommendations for improving the systems.
A group of experts has developed a framework to research and apply brain-cell-based computer technology responsibly. The technology has significant promise for accelerating disease understanding but raises questions about the moral status of bio-computers.
Researchers will develop new ecological models to analyze data from camera traps, satellite images, and DNA sequences to monitor biodiversity changes around the world. The project aims to detect early warnings of species moving north due to climate warming, providing insights into mechanisms of negative impacts on biodiversity.
A new UTHealth Houston study reveals that implementing artificial intelligence-powered software can significantly reduce treatment time for stroke patients undergoing endovascular thrombectomy, with an average reduction of 11 minutes in time to thrombectomy initiation.
Researchers at Osaka University develop a method to train AI models using simulated city images, reducing the need for real data and saving human effort. The approach generates realistic images with accurate ground truth labels, addressing instance segmentation challenges.
A study published in Scientific Reports found that large language model AI chatbots can generate creative ideas at least as well as the average human in a divergent thinking task. However, human participants with the highest scores still outperformed the best chatbot responses, especially in categories such as fluency and flexibility.
An AI system by Lunit identified improperly positioned endotracheal tubes with high sensitivity and specificity, allowing for earlier repositioning and reducing complications. The study included 539 chest radiographs from three institutions and showed promising results in detecting ETT presence and improper position.
Researchers found that even top-performing AI models can mistake nonsense sentences for natural language, highlighting the need for improvement. The study suggests that comparing human language processing to AI's could provide new perspectives on how we think.
Researchers developed an AI foundation model for eye care that can identify sight-threatening eye diseases and predict general health. The model, RETFound, was trained on millions of eye scans from the NHS and outperforms existing state-of-the-art AI systems across complex clinical tasks.
A pilot study found that ChatGPT performed as well as doctors in suggesting the most likely diagnoses for patients being assessed in emergency medicine departments. The chatbot generated a list of likely diagnoses and suggested the most likely option, with a large overlap between its shortlist and those of doctors.
A new paper suggests that ecology can inspire more powerful, resilient, and socially responsible AI to solve complex problems. Artificial intelligence systems can incorporate greater amounts of data and diverse sources, discovering new interactions and drivers that may not have been considered before.
Researchers developed a method to predict rice yield using smartphone photos, showing 68% accuracy and comparable results to satellite data. The model's simplicity makes it suitable for mobile devices and can be easily transferred to other crops.
A study found that people perceive AI-assisted messages as less effortful and more improper than manually written messages. As a result, participants expressed lower satisfaction with their relationship and increased uncertainty about the friendship. Additionally, receiving an AI-aided response also led to negative perceptions of Taylo...
The article discusses how artificial intelligence can enhance climate modeling by generating large ensembles of simulations at moderately high resolutions. This approach capitalizes on advances in computing and AI to learn from data, enabling more accurate predictions and reduced uncertainty. By focusing on resolutions between 10-50 km...
A new study by USF researchers found that linguistics experts were unable to distinguish between AI- and human-generated abstracts less than 39 percent of the time. The study concluded that ChatGPT can write short genres as well as humans, while longer texts are easier to identify as AI-generated
Scientists are using AI to analyze pollen grains in sediments and air samples to understand past environmental change. The new system can classify over a thousand pollen grains in under an hour, providing insights into biodiversity and climate change.
A new AI tool assesses motor performance using finger taps, providing rapid and standardized ratings, comparable to expert neurologists. The test outperformed primary care physicians in accuracy, opening doors to evaluating other movement disorders.
A novel AI system developed by City University of Hong Kong improves predictive accuracy in dense traffic, reducing latency and increasing efficiency. QCNet achieves speed and accuracy in predicting road users' movements, even with long-term predictions, making autonomous driving safer and more human-like.
Prof. Daniel Trefler, a renowned economist at the University of Toronto's Rotman School of Management, has been elected a fellow of the Royal Society of Canada. He is recognized for his outstanding contributions to research on international trade, innovation, inequality, and artificial intelligence.
Researchers found that an AI algorithm comparable to human readers demonstrated high sensitivity and specificity in detecting breast cancer. The study utilized a large dataset of mammographic exams and compared the performance of AI with human readers, finding no significant difference between the two.
A recent study found that ChatGPT scored 9/10 in accuracy for debunking Covid-19 vaccine myths, making it a reliable source of non-technical information. The AI tool's responses were accurate in addressing both true and false contraindications, but highlighted potential concerns over changes in answers over time.
Researchers at Auburn University are developing a Findable, Interoperable, Accessible, and Reusable (FAIR) data platform to manage fusion device data according to FAIR standards. The project aims to accelerate fusion energy research by enabling strong collaborations and promoting diversity in the workforce.
Researchers from Helmholtz Munich and TU Dresden develop AI-powered biomarker detection that outperforms previous approaches. The algorithm achieves high sensitivity on resection tissue samples and shows promise as a pre-screening tool for biopsy tissue.
A new study reveals that people judge smart AI assistants as partly responsible for decisions, even when they're used as tools. Participants saw them as more responsible for positive outcomes, highlighting the need for organizations to consider social and moral norms in AI design.
A deep learning approach has unveiled a significant change in the characteristics of global daily precipitation for the first time. The research found that on more than 50% of all days, there was a clear deviation from natural variability in the daily precipitation pattern since 2015.
A team of researchers from the University of Zurich and Intel has developed an AI system called Swift that can beat human champions in drone racing. The autonomous drone achieved the fastest lap, winning multiple races against three world-class champions, but human pilots proved more adaptable to changing conditions.
Researchers created a robot-assisted dressing system using artificial intelligence to accommodate various human body shapes, arm poses, and clothing selections. The system was evaluated in a human study with 510 trials across 17 participants, achieving an average success rate of 86% in pulling the sleeve of each garment onto their arm.
A study published in Radiology found that combining short- and long-term breast cancer risk models using artificial intelligence can improve cancer risk assessment. The combined model showed an overall improved risk assessment for both interval and long-term cancer detection, identifying women at high risk for breast cancer.
Researchers investigated the quality of CPR directions provided by AI voice assistants and found they lacked relevance and consistency. The study recommends relying on emergency services for life-saving care instead of self-directed guidance from AI tools.
Researchers from Radboud University have successfully decoded brain signals into intelligible speech using a combination of implants and advanced artificial intelligence models. The technology has been tested with an accuracy of 92-100%, showing promise for individuals with locked-in states who are paralyzed and unable to communicate.
Two studies demonstrate that artificial intelligence in colonoscopy can increase adenoma detection rates and reduce miss rates, but also lead to increased removal of non-neoplastic polyps. The findings suggest the need for additional research on more defined detection parameters before AI can be integrated into routine clinical care.
Researchers at George Mason University are developing a new biometric technology that uses sweat analysis to verify identities. The project, led by Emanuela Marasco, aims to characterize a person's extrinsic and intrinsic features for more accurate representation of their identity.
Researchers Elzbieta Izbicka and Robert T. Streeper discuss the evolution of cancer drug development, highlighting recent progress and emerging challenges due to environmental factors. They also examine how artificial intelligence may impact future drug development.
The WVU team evaluated Code Interpreter's features, finding it accessible to students but limited for scientists working with biological data. The plugin breaks down barriers for coding, but lacks internet access and parallel processing capabilities.
A new study found that younger atrial fibrillation (AFib) patients who underwent MRI-guided ablation treatments had better outcomes. The treatment significantly improved arrhythmia recurrence rates by 50% for patients under 58 years old.
Researchers found AI chatbot responses to top cancer queries are generally accurate but lack actionable information. These limitations suggest relying on chatbots for supplementary purposes rather than primary medical information.
A study published in Scientific Reports found that ChatGPT achieved similar or higher average grades than university students in 9 of 32 courses, with notable strengths in 'Introduction to Public Policy'. The survey also revealed that 74% of students would use ChatGPT for assignments, while educators underestimated this trend
Researchers from Brigham and Women's Hospital evaluated ChatGPT's recommendations for cancer treatment aligned with National Comprehensive Cancer Network (NCCN) guidelines. They found that approximately one-third of the chatbot's responses were non-concordant, highlighting its limitations in clinical decision-making.
The NIMS-OS middleware coordinates materials-search AI and robotic experiment systems for efficient automation. The system successfully executed a model experiment to identify effective electrolytes for lithium metal electrodes.
Researchers at MIT developed a method to simplify the process of whole-body manipulation for robots, enabling them to reason efficiently about moving objects. The technique uses AI and smoothing to reduce the number of decisions required, making it possible for robots to adapt quickly in complex environments.
Researchers at UCSF and UC Berkeley have developed a brain-computer interface (BCI) that allows a woman with severe paralysis from a brainstem stroke to speak through a digital avatar. The system can decode brain signals into text at nearly 80 words per minute, making it a vast improvement over commercially available technology.
Researchers found that subscription channels have greater source credibility and content control, leading to higher conversion rates despite perceived ad intrusiveness. In contrast, AI-recommended channels elicit more clicks but with smaller proportions of genuine interest and purchases.
Researchers from Okayama University used CNNs to estimate rice yield from pre-harvest photographs, achieving accuracy of 68-69%. The model highlighted the importance of panicles in yield estimation and showed promise for monitoring rice productivity at regional scales.
The University of Missouri is launching a five-year, $3 million doctoral training program to prepare the next generation of scientists and engineers for emerging fields like materials science and data science. The program aims to empower future workers with both technical expertise and data-driven insights.
A study found that a chatbot could generate responses comparable to those of ophthalmologists, without significantly more incorrect information or harm potential. The results suggest that AI can provide adequate support for patients seeking eye care advice, with minimal risk of adverse effects.
A new study found ChatGPT to be nearly 72 percent accurate across all medical specialties and phases of clinical care. It was also 77 percent accurate in making final diagnoses. However, the model struggled with differential diagnosis, which is a crucial aspect of medicine.
A study published in Neurology has identified markers of Parkinson's disease in eye scans, detecting the condition up to seven years before clinical presentation. Researchers used AI-powered analysis of retinal images from two large datasets to uncover subtle changes associated with Parkinson's.
Biased AI can limit climate predictions and misguide governments due to missing information from under-represented communities. Human-in-the-loop design can fill these 'data holes' by offering a sense check on used data and context.
Researchers at Brigham and Women's Hospital have developed an AI model that screens electrocardiograms (ECG) for signs of atrial septal defects (ASD), a common adult congenital heart disease. The model correctly detected ASD in 93.7% of cases, outperforming traditional methods.
The researcher aims to bridge completeness, efficiency, and applications in 3D graphs to solve problems in physics, fluid dynamics, and biotechnology. Geometric graphs can represent molecules, proteins, and drugs, enabling the prediction of their behavior and properties.
A new study by University of East Anglia reveals ChatGPT's systematic left-wing bias, favoring Democrats in the US and Labour Party in the UK. The platform's responses also lean towards President Lula da Silva of Brazil's Workers' Party.
The University of Missouri is using a $1 million grant to develop an Industry 4.0 lab, providing engineering students with hands-on learning experiences in the latest industrial revolution's technology-centered job market. The lab will integrate skills at a higher level and keep students at the state-of-the-art level for industry.