Recent research has unveiled two promising approaches that could dramatically reduce the costs of running large language models (LLMs) while simultaneously improving their performance on complex reasoning tasks. These innovations come at a critical time as enterprises increasingly deploy AI solutions but struggle with computational expenses. Chain of draft: Less is more Researchers at Zoom … Read more
The Association for Computing Machinery has awarded the prestigious Turing Award to Andrew Barto and Richard Sutton for their groundbreaking work on reinforcement learning. As reported by Cade Metz in The New York Times, the two researchers will share the $1 million prize that comes with what is often called the “Nobel Prize of computing.” … Read more
Inception Labs has unveiled Mercury, a new family of diffusion-based large language models (dLLMs) that can generate text up to 10 times faster than conventional autoregressive LLMs. According to the company, Mercury models can process over 1,000 tokens per second on NVIDIA H100 GPUs, speeds previously achievable only with specialized hardware. The company’s first publicly … Read more
You.com has unveiled a new AI research tool called Advanced Research & Insights agent (ARI) that can analyze more than 400 sources at once. According to CEO Richard Socher, interviewed by Michael Nuñez for VentureBeat, the tool aims to transform market research by producing comprehensive reports in minutes instead of weeks. ARI features direct source … Read more
Hugging Face has developed an open-source version of autonomous research technology, matching key capabilities of OpenAI’s recently launched Deep Research feature. As reported by Benj Edwards for Ars Technica, the project called “Open Deep Research” was completed within 24 hours of OpenAI’s announcement. The new tool enables AI models to independently browse the web and … Read more
Researchers from Stanford and the University of Washington have developed an AI model called s1 that rivals the capabilities of expensive commercial AI systems while costing less than $50 in computing resources to train. The model, which was created through a process called distillation using Google’s Gemini 2.0 Flash Thinking Experimental model, demonstrates similar performance … Read more
Meta AI and the University of Illinois Chicago researchers have created a new system that helps AI models respond more efficiently to queries. As reported by Ben Dickson in VentureBeat, the technology enables AI to determine whether a question needs deep reasoning or a quick response. The system, called Inference Budget-Constrained Policy Optimization (IBPO), uses … Read more
OpenAI has introduced a new AI agent called “deep research” that enables ChatGPT to conduct comprehensive research tasks. The feature, currently available to ChatGPT Pro users with limited queries per month, utilizes OpenAI’s o3 reasoning model to analyze information from multiple web sources and compile detailed reports with citations. The system can process text, PDFs, … Read more
OpenAI has formed a partnership with U.S. National Laboratories to apply its AI models to scientific research and nuclear security initiatives. According to Hayden Field at CNBC, the collaboration will give up to 15,000 scientists access to OpenAI’s o1 series models. The partnership aims to enhance power grid cybersecurity, advance disease treatment research, and deepen … Read more
Researchers from the Center for AI Safety and Scale AI have unveiled “Humanity’s Last Exam,” which they assert is the most challenging test ever created for artificial intelligence systems. According to reporting by The New York Times, the test consists of approximately 3,000 multiple-choice and short-answer questions spanning fields from analytic philosophy to rocket engineering. … Read more