Researchers from UC San Diego and Tsinghua University have developed a method that improves artificial intelligence’s ability to understand when to use external tools versus relying on built-in knowledge, similar to how human experts approach problem-solving. Using a relatively small language model with 8 billion parameters, the team achieved a 28% improvement in answer accuracy …
The rise of AI could lead to a 3-12% increase in global electronic waste by 2030, amounting to an extra 2.5 million metric tons annually, according to a study by researchers at the Chinese Academy of Sciences and Reichman University in Israel published in the journal Nature Computational Science. The analysis, based on industry investment …
The research paper “Mind Your Step (by Step): Chain-of-Thought can Reduce Performance on Tasks where Thinking Makes Humans Worse” investigates the effectiveness of chain-of-thought (CoT) prompting in large language and multimodal models. While CoT has generally improved model performance on various tasks, the authors explore scenarios where it may actually hinder performance, drawing parallels from …
A new study by researchers from Technion, Google Research, and Apple reveals that large language models (LLMs) have a deeper understanding of truthfulness than previously thought. The researchers analyzed the internal representations of LLMs across various datasets and found that truthfulness information is concentrated in specific response tokens, VentureBeat reports. By training classifier models on …
Researchers at XJDR have developed a new technique called Entropix that aims to improve reasoning in language models by making smarter decisions when the model is uncertain, according to a recent blog post by Thariq Shihipar. The method uses adaptive sampling based on two metrics, entropy and varentropy, which measure the uncertainty in the model’s …
DeepMind researchers have introduced a new agentic framework called Talker-Reasoner, which is inspired by the “two systems” model of human cognition. The framework divides the AI agent into two distinct modules, VentureBeat reports: the Talker, which handles real-time interactions with the user and the environment, and the Reasoner, which performs complex reasoning and planning. The …
OpenAI has developed a new AI model that can generate media content such as images, videos and audio 50 times faster than previous systems. The new model, called a “continuous-time consistency model,” takes about a tenth of a second to generate an image instead of the usual five seconds, OpenAI researchers Cheng Lu and Yang …
The research company Playground Research presents “Playground v3”, a new AI model for text-image generation, which has apparently achieved top performance in several test procedures. The system stands out for its precise implementation of text instructions, its ability to reason logically, and the outstanding quality of its text rendering. In user studies, the model even …
Researchers have developed a new method called “linear-complexity multiplication” (ℒ-Mul) to make calculations in artificial intelligence more efficient. The method replaces complex multiplications with simpler additions, according to Jason Hickey and his team at the Google AI Research Center in Accra. The researchers showed that ℒ-Mul achieves the same accuracy as traditional methods for language …
Microsoft and Tsinghua University have developed a new AI architecture called “Differential Transformer” that improves the performance of large language models. Furu Wei from Microsoft Research told VentureBeat that the new method amplifies attention to relevant contexts and filters out noise. This is designed to reduce problems such as the “lost-in-the-middle” phenomenon and hallucinations in …