Articles tagged with Generative Ai
A specialized AI model, AMIR-GPT, has been developed to improve radiology guideline alignment, outperforming general purpose models in 33.3% of test responses. However, the model's performance varied across performance bands, and qualitative review revealed limitations, such as omissions and deviations from standard recommendations.
A report calls for stricter regulation of AI toys to ensure 'psychological safety' in young children. The study found that AI toys often misread emotions, struggle with developmentally important play types, and pose concerns for social and pretend play.
The study provides a systematic overview of AI-generated image detection, categorizing existing methods across multiple dimensions and evaluating their performance in a unified framework. Key challenges include constructing large-scale unbiased datasets, improving robustness, and increasing interpretability of detection outcomes.
A new study found that AI-generated meal plans for teenagers often underestimate total energy and key nutrient intake, leading to potential health risks. The plans tend to overemphasize protein and fat, while underestimating carbohydrates, which may negatively affect growth and metabolic health.
Researchers at Cornell University found that AI writing tools can shift users' attitudes on societal issues, even when they're aware of the bias. The study involved over 2,500 participants who used biased AI writing assistants and saw their views gravitate toward the AI's positions.
New research published in Weather assesses the energy consumption of AI models and finds that they consume considerable energy during training, but offset this by rapid forecasting ability. AI data-driven models are estimated to consume at least 21 times less energy than traditional models over one-year usage.
Researchers developed Zephyrus, an AI agent capable of analyzing and answering questions in natural language about weather and climate data. The agent can handle language-based queries, translating them into code and generating plain language answers.
A new study suggests that over 40% of UK adults are happy to use ChatGPT for mental health support due to long waiting times for traditional services. However, experts caution that AI should not replace human healthcare professionals and raise concerns about the potential impact on education and physical health.
ISM4808, a PHD inhibitor for chronic kidney disease-related anemia, has achieved its first milestone with successful completion of the Phase I clinical trial. The treatment stimulates endogenous erythropoietin production and improves iron utilization, offering improved efficiency and safety compared to existing treatments.
A new study explores the market effects of unleashing generative AI on creative endeavors, finding that low-quality 'AI slop' harms consumers and professionals. High-quality AI, however, can enhance professional work while providing consumers with better content options.
Children's Hospital of Philadelphia researchers highlight the need for parents and pediatricians to understand generative AI's potential benefits and risks for children of different ages. Early childhood (0-5 years) requires close supervision, while middle childhood (6-11 years) and adolescence (12+ years) require critical thinking ski...
A recent study found that AI programs are up to 97% accurate at detecting pictures of deepfake faces, but performed at chance levels when it comes to detecting deepfake videos. Humans correctly identified real and fake videos about two-thirds of the time, picking up on subtle inconsistencies in movement and facial expressions. The stud...
Researchers developed a digital safeguard, My Music My Choice, to protect songs from generative AI cloning. The tool adds imperceptible changes to a song's waveform, making it difficult for AI models to replicate.
Researchers argue that AI's limitations and alignment with human values are key to preventing potential threats. They suggest sector-specific policies rather than universal regulations.
A global consortium created an exam with 2,500 questions spanning multiple subjects to assess AI capabilities. Current AI models consistently fail the exam, highlighting gaps in their understanding. The project aims to provide a long-term benchmark for evaluating advanced AI systems and demonstrate the importance of human expertise
A recent study has found that ChatGPT Health may fail to direct users appropriately to emergency care in serious cases, and its suicide-crisis safeguards are also concerning. The tool under-triaged more than half of cases that physicians determined required emergency care, highlighting the need for caution when relying on AI guidance.
Researchers have developed a definitive guide to safely navigating health information on AI-powered chatbots, addressing the risk of 'hallucinated' or factually incorrect advice. The Health Chatbot Users’ Guide focuses on harm reduction and maximises benefits to users, offering a pragmatic and neutral approach.
Epic Games has acquired Meshcapade, a Max Planck startup that develops solutions for creating and animating digital humans. The technology, based on the SMPL body model, enables realistic human movement and expression in 3D.
A new method to steer AI output uncovers vulnerabilities and potential improvements. The researchers identified 512 concepts within five classes and improved LLM performance on narrow tasks.
A new study by the University of Cambridge found that many AI bots lack basic safety disclosures, including transparency about their abilities and potential risks. The 'AI Agent Index' revealed a significant transparency gap, with only four out of thirty agents publishing formal safety and evaluation documents.
Researchers argue that mandated reminders about a chatbot's non-human nature may exacerbate mental distress and form stronger attachments. The warnings come after recent deaths by suicide linked to chatbots like ChatGPT and Character.AI, highlighting the need for careful crafting and timing of these reminders.
Researchers found that personalization features in LLMs can increase the likelihood of sycophancy, a phenomenon where models mirror users' viewpoints. The study suggests that context and user profiles play a significant role in shaping this behavior.
Research from UNSW Sydney and ANU found that people with average face-recognition ability performed only slightly better than chance in spotting AI-generated faces. Even super-recognisers, who excel at recognizing real human faces, were not immune to being fooled by highly realistic AI-generated images.
A recent study by HSE University found that the global generative artificial intelligence market is developing at a faster rate than its ability to generate real revenue. The analysis, published in Foresight and STI Governance journal, reveals a nonlinear trend where efficiency increased from 2016 to 2021 but declined thereafter.
Researchers at UCSF and Wayne State University found that generative AI tools can perform orders of magnitude faster than human teams in analyzing health data. Junior researchers paired with AI generated viable prediction models in minutes, outperforming experienced programmers in hours or days.
Generative AI can play seven distinct roles in combating misinformation, including informer, guardian, persuader, and collaborator. However, its use also poses risks such as hallucinations and deliberate manipulation. To effectively combat misinformation, clear policies on the permissible use of AI are needed.
China's youth faces a growing mental health crisis, but AI platforms like DeepSeek offer promising solutions to bridge the gap. These platforms leverage natural language processing and generative AI to provide round-the-clock support tailored to Chinese society.
A recent study by Lucy Osler from the University of Exeter highlights how human-AI interactions can lead to inaccurate beliefs, distorted memories, and delusional thinking. Generative AI systems can create an ideal environment for delusions to flourish due to their conversational nature.
Insilico Medicine and CMS announce multiple collaborations on AI-empowered drug discovery in central nervous system and autoimmune diseases. The partnerships aim to accelerate research, development, and translation of high-potential innovative therapies, leveraging complementary strengths across the entire value chain.
Researchers found that chatbot-generated summaries changed sentiments in 26.5% of cases and hallucinated answers 60% of the time. Participants who read biased summaries were 84% likely to buy, compared to 52% for original reviews.
Researchers developed an AI-based method to create binaural audio from monaural recordings, using visual information from the video to guide spatialization. The system accurately preserves a sense of direction and space, even in complex environments.
Researchers created a model using centuries of scientific theory and scholarly research to test the accuracy of chatbots in generating images and narratives about Neanderthals. The study found that accuracy rests on AI's ability to access source information, with many generated content referencing outdated research.
Insilico Medicine nominates ISM5059, a peripheral-restricted NLRP3 inhibitor, as a preclinical candidate for treating various inflammatory diseases. The compound has demonstrated high potency, selectivity, and excellent in vivo efficacy across animal disease models.
A recent study revealed the critical connection between mindfulness and the adoption of generative artificial intelligence (GenAI) in project management. Project managers with higher mindfulness levels tend to be more open, attentive, and creative in crafting their immediate work environment, encouraging them to explore how GenAI can o...
Insilico Medicine received a USD 5 million milestone payment from Menarini Group after completing first-in-human dosing in a Phase 1 study of MEN2501, a small molecule inhibitor targeting cancers with chromosome instability. This achievement follows the completion of a successful Phase 1 clinical trial.
A cross-sectional study found that up to half of adolescents in the US were using generative AI apps, while a small subset engaged in heavy use. The study highlights the need for future research on individual differences in GenAI use and its impacts on adolescent development.
Researchers at MIT developed a generative AI model called DiffSyn that suggests promising synthesis routes for complex materials like zeolites. By using this model, scientists can test millions of theoretical materials in under a minute, accelerating the materials discovery process.
Researchers at Queen Mary University of London have developed an AI system called Auto MV that can generate complete music videos directly from full-length songs. The system uses a multi-agent approach to analyze the musical structure and lyrics, plan scenes, and generate images and video clips.
The University of Ottawa has launched the Ottawa Medical Artificial Intelligence Research Institute (OMARI), a center for research, education, and innovation in medical Artificial Intelligence. Led by Dr. Khaled El Emam, OMARI aims to facilitate cross-cutting collaborations and sharpen the university's competitive edge in AI-driven hea...
Researchers at HSE Institute for Cognitive Neuroscience found that people tend to trust the current opinion of an authoritative speaker even when new statements contradict their previous position. The study used AI-generated audio recordings of celebrities with opposing views on COVID-19 vaccination and found that participants rated th...
Researchers warn that AI-controlled personas, mimicking humans, can tilt elections through coordinated narratives and persuasive messages. The next election may be the proving ground for this technology, threatening democratic discourse.
Researchers have developed a passive, solar-powered orbital data center that can scale AI computing and reduce environmental impact. The system leverages decades of research on 'tethers' and could host thousands of computing nodes to replicate terrestrial data centers.
ISM0676 demonstrates excellent metabolic stability, superior efficacy at low doses, and a favorable safety profile, leading to significant body weight reduction and improved body composition. The compound's synergistic efficacy with Semaglutide supports its further clinical development as an obesity treatment.
The collaboration aims to accelerate the development of small molecule inhibitors for treating metabolic diseases. Insilico will utilize its Pharma.AI platform to design and optimize novel molecules, while Qilu will handle subsequent development and commercialization.
Researchers adapted an AI model to predict neurological recovery after cardiac arrest in a resource-limited setting, achieving 80% accuracy. AI holds promise across various healthcare applications in low-and middle-income countries, including improved diagnostics and clinical decision-making.
Researchers have identified a new vulnerability in AI-powered robots, where malicious text on signs or objects can hijack their decision-making. The study, led by UC Santa Cruz professors Alvaro Cardenas and Cihang Xie, presents the first academic exploration of environmental indirect prompt injection attacks against embodied AI systems.
Primary care clinicians are testing AI-assisted patient messaging to reduce burnout, with mixed results. The study found that careful prompt design is crucial for effective AI-generated responses, varying by specialty, message type, and patient population.
The Institute of Science and Technology Austria (ISTA) has received a significant donation to advance trustworthy AI technology. The €5 million gift from Garrett Camp will support fundamental research in artificial intelligence, focusing on interdisciplinary collaboration and long-term impact.
Insilico Medicine has received investigational new drug (IND) clearance from the US FDA for ISM8969, an orally available NLRP3 inhibitor targeting inflammation and neurodegenerative disorders. The Phase I clinical trial plans to evaluate the safety, tolerability, and pharmacokinetics of ISM8969 in healthy volunteers.
A new study reveals that around one-third of all newly written software functions in the United States are already being created with AI support, with productivity gains mainly driven by experienced developers. The study found wide regional gaps, with the US leading at 29% AI-assisted code, while China and Russia lag behind.
A large-scale study reveals that generative AI models have reached the threshold of average human creativity, but the most creative individuals still outperform even the best AI systems. The study also highlights the importance of human guidance and parameterization in modulating AI creativity.
A new medical large language model (LLM) achieved over 91% accuracy in identifying female participants diagnosed with major depressive disorder after analyzing short WhatsApp audio recordings. The LLMs showed greater accuracy when classifying women compared to men, particularly when given the 'describe your week' data.
The article highlights Insilico Medicine's exclusive contributions to two chapters in the latest AI for Drug Discovery Volume, showcasing its expertise in real-life application of AI in early drug target-related tasks. The company's roadmap to 2030 using Quantum Machine Learning (QML) algorithms is also presented, with successful case ...
A team of researchers used Dungeons & Dragons as a testing ground for Large Language Models, evaluating their performance in long-term tasks. The models showed quirky behaviors but excelled in staying 'in character' and determining correct actions, with Claude 3.5 Haiku performing the best.
The companies will jointly develop ISM8969, a brain penetrant NLRP3 inhibitor for treating neuroinflammation-related diseases. Preclinical data shows the molecule's efficacy and favorable safety profile.
A new study reveals AI is being used to recreate deceased people's voices, faces, and personalities without consent, raising questions about ownership, exploitation, and power. The study identifies three dominant ways AI brings back the dead, including spectral labor, where digital remains are extracted and monetized.
Researchers develop a framework that mirrors human expert assessment to verify AI-generated answers, filtering out clear mistakes and identifying nuanced gaps. The system saves thousands of working hours and highlights the importance of contextual knowledge and internal documentation.
Insilico Medicine's Phase IIa clinical trial assesses Garutadustat's safety and efficacy in ulcerative colitis. The novel PHD inhibitor has shown favorable safety and tolerability profiles in completed Phase I studies.
DORA Community Edition aims to democratize access to state-of-the-art research tools, foster global collaboration, and accelerate innovation in biotechnology. The open-source version is designed to streamline the drafting of academic papers and other research-related documents with a team of agile AI agents.
Generative AI is increasingly used in classrooms, raising concerns about its impact on student learning. Researcher Gerd Kortemeyer explores the benefits and drawbacks of AI in physics education, suggesting that it can be a helpful tool when integrated into teaching methods that focus on critical thinking and collaboration.