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 81–100
A cascade of new research papers published today on arXiv CS. AI reveals a significant leap in artificial intelligence, particularly across computer vision and image processing....
Read article →For too long, the brilliant, complex minds of AI models have operated behind a veil, their internal logic a mystery even to their creators. But today, the academic world delivers a potent shot of optimism for founders everywhere: two significant research papers published on arXiv...
Read article →The relentless pursuit of efficiency in deploying large AI models has just taken a critical turn, with new research simultaneously revealing a promising avenue for accelerating Large Reasoning Models (LRMs) and issuing a stark warning about the hidden costs of a popular optimizat...
Read article →Nintendo rarely blinks first in a pricing standoff, but Metroid Prime 4: Beyond, a flagship title revered for its ambitious scope, has just received its inaugural substantial discount. This isn't merely a price drop; it's a strategic maneuver from a company known for fiercely gua...
Read article →The venture capital landscape continues to churn with targeted bets and massive scale-ups, evident today as Meridian Ventures announced the close of its $35 million second fund, dedicated to pre-seed and seed-stage companies founded by MBA-deferred entrepreneurs TechCrunch. Simul...
Read article →Meridian Ventures, the firm spearheaded by Devon Gethers and Karlton Haney, has just announced the close of its second fund, securing $35 million to specifically back pre-seed and seed-stage companies founded by individuals who have deferred their MBA studies. This raise, reveale...
Read article →A new wave of AI research, spotlighting architectures like CSI-JEPA and Mini-JEPA, is poised to reshape how foundation models are built and deployed, moving away from monolithic, data-hungry giants towards more specialized, accessible, and resource-efficient solutions. Published ...
Read article →Two new research papers, quietly published on arXiv today, are laying critical groundwork for the next generation of embodied AI, tackling the deeply complex challenges of human-like motion synthesis and robust state estimation in dynamic environments. These breakthroughs, Anchor...
Read article →A significant new research paper, "GFMate: Empowering Graph Foundation Models with Test-time Prompt Tuning," has been published on arXiv, introducing a novel approach that could profoundly enhance the capabilities of Graph Foundation Models (GFMs) arXiv CS. LG....
Read article →The foundational pillars of artificial intelligence are being subtly, yet profoundly, reshaped. Today, May 15, 2026, five distinct research papers hit the arXiv pre-print server, unveiling critical advancements in transformer architectures that promise to address long-standing ef...
Read article →A torrent of new research, released en masse on May 15, 2026, reveals an accelerating, almost relentless push by scientists and engineers to integrate advanced AI and machine learning across virtually every hard science and engineering discipline. This deluge of papers from arXiv...
Read article →A torrent of cutting-edge research, freshly published on arXiv, signals a pivotal moment for AI developers and startups. These papers, all released on May 15, 2026, collectively point towards a future where large language models (LLMs) and complex AI systems are not just powerful...
Read article →A significant leadership challenge is unfolding across Elon Musk’s ambitious ventures, with his newly merged SpaceXAI reportedly losing over 50 employees since February. This brain drain comes to light just as the contentious “Musk v....
Read article →The battlefield for developer tools just got a seismic shift: OpenAI has pushed its powerful Codex AI, a desktop tool designed for writing code and interacting with applications, directly into the ChatGPT mobile app. This move isn't just an update; it's a strategic strike, empowe...
Read article →Atech, a hardware company aiming to bring 'vibe coding' to its products, has successfully closed an $800,000 pre-seed funding round. This significant early-stage investment includes backing from the scout funds of venture giants Andreessen Horowitz (a16z) and Sequoia, alongside N...
Read article →The gaming industry is witnessing a twin surge of innovation this week, with Unknown Worlds' Subnautica 2 exploding onto Steam charts and leaked images revealing Microsoft's strategic new Xbox Cloud Gaming controller. These developments underscore a dynamic ecosystem where both g...
Read article →A new startup project, AI IQ, has ignited a fresh wave of contention across the tech ecosystem this week, launching a platform that assigns estimated intelligence quotients to over 50 of the world's most powerful language models, plotting them on a standard human IQ bell curve Ve...
Read article →A significant leap forward in AI capabilities has emerged with two concurrent research breakthroughs, poised to fundamentally reshape how Vision-Language Models (VLMs) are evaluated and understood. Today, new academic papers highlight both a critical benchmark for evaluating comp...
Read article →New research published on arXiv reveals significant strides in making large language models (LLMs) more efficient and accessible, a critical breakthrough for startups fighting for innovation in a resource-intensive field. Two separate papers, both published on May 14, 2026, explo...
Read article →A trio of groundbreaking research papers, all published today on arXiv CS. AI, are setting the stage for a dramatic leap in how AI understands and generates images and video....
Read article →