Articles tagged with Generative Ai
The collaboration utilized Insilico's generative biology AI platform, Pharma.AI, to identify a novel lead for treating oncology diseases. The joint R&D team addressed the druggability challenge of highly novel targets using novel scaffolds generated by Chemistry42.
Researchers found that AI doctors who recall patients' social information, with privacy control offered, increase patient satisfaction. Human doctors don't require social or medical information to establish a close relationship with patients.
A new study by University of Technology Sydney researchers challenges the Infinite Monkey Theorem, finding it unlikely a monkey can type out Shakespeare's works before the universe ends. With around a 5% chance, even an increased number of monkeys would not produce the Bard's entire works.
A study of 50 U.S.-licensed physicians found that GPT-4 did not significantly improve clinical reasoning compared to conventional resources. The integration of GPT-4 as a diagnostic aid alongside clinicians showed promising results but required further exploration to understand its potential benefits.
Researchers at UC Berkeley found that humans create more diverse and nuanced stories than generative AI tools like ChatGPT. While AI shows improvement, it still struggles with complexity and clichés.
MIT researchers developed a versatile technique that combines diverse data from various sources into a shared language for generative AI models. This approach outperformed traditional techniques by 20% in simulation and real-world experiments.
A new study by Penn State researchers suggests that making AI training data diversity information available can shape users' expectations of algorithmic fairness and trust. Displaying racial diversity cues in AI interfaces can enhance users' perceptions of algorithmic fairness and trust, according to the study's findings.
A recent study reveals that when people are told a story was written by AI, they rate it poorly, suggesting a distrust of AI-generated art. Researchers believe this could hinder the use of AI in public health campaigns to encourage healthy behaviors.
Researchers at University of Zurich used AI to analyze antibiotic resistance using GPT-4 model, creating EUCAST-GPT-expert system for accurate interpretation of antimicrobial resistance mechanisms. The AI system performed well in detecting certain types of resistance but had limitations, while human experts were more accurate but slower.
A new study reveals that AI-driven chatbots may perpetuate racial and ethnic biases in pain assessment, leading to further inequalities in healthcare. Researchers found that Black patients were consistently underassessed for their pain compared to white patients, regardless of whether the rater was human or AI.
A study from Lehigh University and Seattle University found that making patients aware of biases in human healthcare decisions increases receptiveness to AI recommendations. By highlighting the limitations of human judgment, healthcare providers can create a more balanced relationship between patients and emerging technologies.
Researchers found that AI-powered chatbot answers on top 50 most prescribed drugs in US were difficult to read and lacked information, posing potential harm to patients. Only 54% of answers aligned with scientific consensus, highlighting the need for caution when relying on these tools.
Researchers developed AI chatbot UroBot to answer complex urology questions with high accuracy, exceeding human urologists. The model justifies its answers based on European Society of Urology guidelines and has been tested on 200 specialist questions, achieving an accuracy rate of 88.4%.
A study from the University of Kansas found that parents trust AI-generated text, like ChatGPT, more than healthcare professionals' content. Parents rated ChatGPT as more trustworthy, accurate, and reliable in many cases, even when there were significant differences in ratings.
The two-year study aims to explore biases in AI systems and develop a 'human-in-the-loop' framework for quality data discovery. It will investigate how humans can be involved as labelers, prompters, and validators to improve data sets and user interfaces.
A new study from UCSF has found that ChatGPT overprescribes in emergency care situations, providing unnecessary x-rays and antibiotics. The AI model was less accurate than resident physicians, with a tendency to err on the side of caution. Researchers emphasize the need for better frameworks to evaluate clinical information before AI c...
The CityUHK team develops a new federated learning technology integrated with research knowledge graphs and large language models to digitally transform STI services. This will create a comprehensive, domain-specific knowledge base and provide intelligent assistant services for researchers.
Recent studies have found that large AI language models are becoming increasingly unreliable, failing to match human expectations on task difficulty. The models excel in complex tasks but struggle with simple ones, leading to a mismatch between human perception and model performance.
Researchers found that AI-generated essays are most similar to those written by male students from higher socioeconomic backgrounds and private schools. The writing styles also tend to be less varied than human-written essays, with AI favoring longer words and affiliations.
A study by Osaka Metropolitan University found that ChatGPT's diagnostic performance for brain tumors was comparable to that of neuroradiologists, with an accuracy rate of 73%. The model's performance varied depending on the type of clinical report written, with higher accuracy when using reports from neuroradiologist writers.
A recent study published in PNAS Nexus found that widespread adoption of large language models like ChatGPT led to a significant decline in user activity on Stack Overflow. The study highlights the impact of ChatGPT on public knowledge sharing and its implications for AI's future.
A new study reveals that AI chatbots like ChatGPT can offer accurate advice on low back pain, including recommended treatments and exercise. However, the systems had limitations, particularly with complex responses and risk factors.
The article discusses the potential benefits and risks of large language models (LLMs) on collective intelligence and proposes recommendations for action. LLMs can increase accessibility and accelerate idea generation, but also pose risks such as undermining motivation to contribute to collective knowledge commons.
A new AI model called Crystalyze can analyze X-ray crystallography data to determine the structure of powdered crystals. The model was trained on a database of over 150,000 materials and successfully predicted structures for over 100 previously unsolved patterns.
A recent review suggests that artificial intelligence can learn by thinking, similar to humans, through processes such as explanation, simulation, analogy, and reasoning. This finding has implications for understanding the similarities and differences between human and artificial cognition, and could lead to improvements in AI systems.
A survey of UK general practitioners reveals that 20% of doctors use generative AI tools like ChatGPT in their practice. The study highlights the potential benefits of AI in reducing administrative burdens and supporting clinical decision-making, but also raises concerns about errors, biases, and patient privacy.
A Cornell University-led research team found that using a technique called 'cultural prompting' reduces bias in large language models like ChatGPT for over 100 countries. The approach involves specifying a cultural identity directly in prompts, adapting to different cultural norms and values.
A recent study found that ChatGPT exhibits cultural values similar to those in English-speaking and Protestant European countries. The model's responses consistently reflected a focus on self-expression values, such as environmental protection and tolerance for diversity.
A fifth of UK GPs are using generative AI in clinical practice, with most using tools for administrative tasks or to support clinical reasoning. However, experts caution against the limitations and risks of these tools, including inaccuracies, biases, and potential harm to patient privacy.
Researchers at Purdue University found that autonomous vehicles can interpret and respond to commands from passengers using large language models like ChatGPT. This technology allows the vehicle to personalize its driving to a passenger's satisfaction and take into consideration traffic rules, road conditions, and weather.
A study by OHSU researchers found that generative AI models like ChatGPT outperform human students in knowledge assessments for a biomedical and health informatics course. The results raise concerns about cheating but also highlight the need to assess student learning beyond basic knowledge.
A new study from Mass General Brigham researchers found that large language models demonstrated no racial or gender discrimination in opioid treatment recommendations. The results suggest that these AI models have the potential to reduce bias and improve health equity in pain management, which is a critical area where disparities exist.
Rice University researchers developed ElasticDiffusion, a method that separates local and global signals to create non-square aspect ratio images without visual imperfections. The new approach can improve consistency and realism in AI-generated images, but still requires significant computational power.
A pathbreaking study finds that conversations with AI models can significantly reduce conspiracy theory beliefs, particularly among those deeply entrenched in their views. The study, involving over 2,000 participants, showed a 20% reduction in belief and an additional 1 in 4 disavowing the conspiracy altogether.
Generative AI is revolutionizing oncological imaging by expanding datasets and improving image quality. This technology enables predictive oncology and personalized cancer screening, offering new hope in the fight against cancer.
The study found that ChatGPT-4 Vision performed well on text-based radiology exam questions but struggled with image-related questions, achieving an accuracy of 65.3%. The model's performance varied across subspecialties and images, with limitations in interpreting certain types of radiologic images.
A machine-learning algorithm called xFakeSci has been developed to detect AI-generated scientific articles, with a success rate of nearly 94%. The tool analyzes word patterns and bigrams to distinguish between real and fake papers, highlighting the need for comprehensive detection methods as AIs become increasingly sophisticated.