MIT researchers found that large language models can understand the visual world and generate complex scenes. By querying LLMs to self-correct code for images, they improved simple drawings and trained a vision system without using visual data.
MIT and Meta researchers develop PlatoNeRF, a computer vision technique using shadows and machine learning to create accurate 3D models of scenes, improving autonomous vehicles and AR/VR efficiency. Combining lidar and AI, PlatoNeRF offers new opportunities for reconstructions and will be presented at the Conference on Computer Vision and Pattern Recognition.
Name entity recognition (NER) extracts entities from text, traditionally requiring fine-tuning. New large language models enable zero-shot NER, like Amazon Bedrock's LLMs, revolutionizing entity extraction.
Africa's potential in AI and green jobs shines, but global support is crucial, says Prof Patrick Verkooijen, new Chancellor of University of Nairobi. With vast natural resources and a young population, Africa's promise is undeniable as it heads towards being home to one in four people globally by 2050.
AI Agent Capabilities Engineering Framework introduces a mental model for designing AI agents based on cognitive and behavioral sciences. The framework categorizes capabilities into Perceiving, Thinking, Doing, and Adapting, aiming to equip AI agents for complex tasks with human-like proficiency.
AI-powered blood test predicts Parkinson's 7 years early, improving treatment research. Test could be available in NHS labs in 2 years.
Nvidia surpasses Microsoft and Apple to become world's most valuable company with $3.34tn market cap, driven by AI chip dominance. Shares soar as chipmaker overtakes Apple, solidifying position in tech market.
McDonald’s ends AI chatbot drive-thru test after wrong order viral videos raise concerns. Over 100 US locations to remove automated order systems.
Proton, the secure email suite, transitions to a nonprofit foundation led by CEO Andy Yen. Foundation includes Sir Tim Berners-Lee, aims to prioritize people over profits.
80% of AI projects fail due to poor use cases or technical knowledge. Gen AI simplifies complexity, helping companies find valuable applications. "Paperclips & Friends" explores AI to tackle increasing customer support demands, highlighting the importance of measuring problem magnitude.
AI is transforming drug discovery by helping scientists find cost-efficient molecules with desired properties. MIT's SPARROW algorithm streamlines synthesis planning, optimizing costs and experimental steps for new medicines, agrichemicals, and specialized materials.
Automating resource management in large language model training optimizes efficiency, enabling focus on experimentation and innovation. AWS Trainium and AWS Batch integration offers scalable, cost-effective deep learning training with streamlined orchestration.
Explaining data normalization and encoding for machine learning algorithms, including Euclidean distance computation. Demonstrates manual and programmatic methods, showcasing a general normalize-and-encode program in C#.
SoftBank is developing AI-powered "emotion-canceling" tech to alter angry customer voices for calmer calls with customer service. The project aims to reduce operator stress, with plans for launch by March 2026, sparking mixed reactions online.
Tech evangelists advocate for AI education amid global school closures, but author Alex Beard warns of growing risks. 850 million children affected.