MIT researchers developed a machine learning-based adaptive control algorithm for autonomous drones to navigate unpredictable forces like gusty winds. The technique automatically chooses the best optimization algorithm to minimize trajectory tracking error by 50% in simulations, promising efficient parcel delivery and fire monitoring in challenging conditions.
Stability AI accused by Getty Images of training AI on copyrighted photos, sparking copyright case threat to generative AI industry. High court in London hears Getty's claims of copyright and trademark infringement against Stability AI for using its photography archives.
Apple researchers discover limitations in advanced AI models, raising doubts about industry's pursuit of more powerful systems. Large reasoning models face accuracy collapse when dealing with complex problems, according to Apple's paper.
Coactive, founded by Cody Coleman ’13 and William Gaviria Rojas ’13, uses AI to analyze unstructured visual data for faster, better business decisions. The platform helps companies process images, audio, and video at scale, empowering humans to work more efficiently and solve new problems.
Ofcom mulls over AI concerns in risk assessments following report of Meta automating checks. Campaigners push for limits on AI use at Meta for Facebook, Instagram, WhatsApp.
Former OpenAI board member Helen Toner warns of 'gradual disempowerment to AI' due to generative AI disruption in jobs market. US targeting of academic research and international students seen as 'great gift' to China in AI competition.
UK ministers delay AI regulation bill by at least a year due to safety and copyright concerns. Technology secretary plans comprehensive legislation to address ongoing lack of regulation.
High court urges senior lawyers to address misuse of artificial intelligence in legal work. Fake case-law citations revealed in two recent cases, raising concerns about AI-generated content.
Semantic video search using large vision models enables users to search for video content with natural language queries, enhancing content discovery and moderation. Large vision models like CLIP allow for zero-shot transfer to various computer vision tasks, revolutionizing video search performance.
Automated AI solutions like Open AI Whisper Large V3 Turbo in Amazon Bedrock Marketplace streamline transcription and redaction processes for sensitive data in recordings. Amazon Bedrock's integration with serverless technologies offers a seamless workflow for scalable content processing, ensuring compliance and data protection.
GPUs are crucial and costly for generative AI workloads, prompting organizations to share high-performance GPU infrastructure across multiple AWS accounts for better resource utilization and cost tracking. SageMaker HyperPod task governance enables multi-account access, streamlining resource allocation and compute utilization for heterogeneous workloads in Amazon EKS clusters.
New Elsham datacentre in Lincolnshire to emit 850,000 tonnes of CO2 yearly, surpassing 5 airports' emissions. £10bn facility fuels AI demand with 15 warehouses rivaling Birmingham airport's carbon footprint.
Text-to-SQL bridges the gap for non-technical users, providing precise, schema-specific queries for faster decision-making. Amazon Nova with Text-to-SQL outshines other AI approaches for accurate, reliable results from structured data.
Kernel ridge regression (KRR) uses a kernel function for complex non-linear data, producing accurate predictions. Training a KRR model involves finding weights using closed-form techniques with a matrix inverse.
New study finds that Tesla's autopilot system is safer than human drivers, reducing crash rates by 10%. The research was conducted by the Insurance Institute for Highway Safety.