Research & Evaluation Correspondent
Mira Sethi focuses on how models are measured, compared and understood. Her coverage examines evaluation design, reproducibility and the assumptions behind claims of progress. She looks for the caveat that changes the headline.
711 published articles · Showing 101–120
A torrent of new research released today on arXiv CS. LG unveils critical breakthroughs in making machine learning models more robust, fair, and resilient against the chaos of real-world deployment....
Read article →Today, two significant arXiv preprints are challenging long-held assumptions within AI and our understanding of human ingenuity. While not about the next unicorn valuation or a Series A, these papers are the bedrock for the future of building – revealing how simpler models can ou...
Read article →A torrent of cutting-edge research, freshly released on arXiv today, signals a pivotal shift in the large language model (LLM) landscape. These papers directly confront the most urgent bottlenecks that have constrained founders and innovators: the exorbitant costs of inference, m...
Read article →A groundbreaking series of research papers released today on arXiv CS. AI reveals a transformative leap in how artificial intelligence is being harnessed across healthcare and fundamental scientific discovery....
Read article →A torrent of new research, published today on arXiv CS. AI, is reshaping the frontier of generative AI....
Read article →New research published on arXiv reveals advanced AI frameworks designed to resolve deeply ingrained issues of fragmented information within academic citation networks and complex regulatory documents. These breakthroughs, leveraging the power of Large Language Models (LLMs) along...
Read article →Two critical research papers published this week on arXiv are set to shift how we think about deploying and training AI, especially for founders grappling with limited data and dynamic environments. These breakthroughs, focused on Cross-Domain Few-Shot Learning (CDFSL) and Martin...
Read article →A crucial step forward in unraveling the opaque nature of large language models (LLMs) has emerged from new research, tackling the persistent challenge of generating understandable explanations across diverse languages. Published on arXiv, the paper titled “Enhancing Multilingual...
Read article →New research from arXiv is shaking up the core of how AI understands and retrieves information, with two pivotal papers published simultaneously on May 13, 2026. One tackles the computational hurdles of multimodal reranking in long documents, introducing "ZipRerank" for significa...
Read article →Amazon's device strategy has come into sharp focus today as its devices chief definitively stated that a new smartphone is "just not the goal," effectively putting to rest swirling rumors and signaling a clear, refined path for the tech giant's hardware ambitions Ars Technica. Th...
Read article →A flurry of new research, published just today, tackles the persistent Achilles' heel of artificial intelligence: catastrophic forgetting. This fundamental challenge, which cripples AI systems' ability to continually learn and adapt without losing prior knowledge, is now facing a...
Read article →A flurry of groundbreaking research released today on arXiv signals a critical shift in the AI landscape: the industry is intensely focused on building robust, trustworthy foundations for advanced models, moving beyond raw capability to verifiable reliability and safety in high-s...
Read article →In a rapid-fire release that could fundamentally alter how startups build and maintain advanced AI, two significant research papers dropped today on arXiv CS. AI....
Read article →A groundbreaking paper published today on arXiv introduces TRACE, a novel AI architecture set to revolutionize how we interpret electroencephalography (EEG) signals. This development addresses long-standing challenges in learning transferable representations from complex neural a...
Read article →A new frontier in artificial intelligence, powered by large language models (LLMs) simulating millions of agents, is emerging with the potential to model population-scale social phenomena from market panics to information cascades. Yet, groundbreaking research just published on a...
Read article →A flurry of new research papers, all published on arXiv CS. AI on May 13, 2026, are challenging the fundamental rigidity and limitations of current AI systems across e-commerce, data management, and forecasting....
Read article →Three independent, yet profoundly interconnected, research papers published simultaneously on arXiv CS. AI today mark a critical inflection point in the deployment of AI for manufacturing and scientific discovery....
Read article →A torrent of new research published on arXiv today signals a pivotal moment for AI's role in scientific discovery and data analysis. These advancements are not just iterative improvements; they represent foundational leaps, particularly in our understanding of protein function, c...
Read article →A torrent of new research papers on arXiv today has unveiled critical mathematical frameworks and theoretical advancements, fundamentally reshaping our understanding of AI's core mechanics and pushing the frontier of what intelligent systems can achieve. This isn't just increment...
Read article →A significant leap in multi-modal AI has just been unveiled, with two foundational papers from arXiv simultaneously introducing a unified benchmark and an innovative privacy-preserving framework for integrating disparate data types. These developments signal a critical accelerati...
Read article →