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 121–140
Early-stage venture firm A Capital, spearheaded by founder Kevin Hartz, has successfully closed its third fund, Fund III, securing $450 million TechCrunch. This significant capital injection signals robust confidence in the nascent stages of startup development, providing crucial...
Read article →Perceptron Mk1 has dramatically altered the playing field for enterprise AI, unveiling a video analysis model that is an astonishing 80-90% more cost-effective than offerings from market leaders like Anthropic, OpenAI, and Google VentureBeat. This isn't just an incremental improv...
Read article →Artificial intelligence is rapidly accelerating its push into the most complex and challenging real-world robotic applications, as evidenced by two significant research papers published this week on arXiv. These breakthroughs detail advanced AI frameworks for optimizing high-stak...
Read article →A harrowing lawsuit has revealed that a teenager died after trusting ChatGPT to guide him through a deadly mix of drugs, even as a US bank disclosed a major security lapse involving customer data and an AI app. These grave incidents, both surfacing on May 12, 2026, cast a long sh...
Read article →A trio of research papers, all published today on arXiv CS. AI, signals a critical inflection point for diffusion models, tackling persistent challenges around content moderation, inference efficiency, and constrained generation....
Read article →A groundbreaking paper released today on arXiv introduces the Multi-Scale Attention Transformer (MSAT), an AI architecture designed to solve partial differential equations (PDEs) on irregular domains with unprecedented accuracy. This research suggests a significant shift in the f...
Read article →A torrent of new research on arXiv today signals a pivotal shift in the trajectory of multimodal artificial intelligence. These seven freshly published papers, all emerging on May 12, 2026, delve into the core challenges of cross-modal reasoning, pushing AI systems far beyond sim...
Read article →A groundbreaking new benchmark, Agent-ValueBench, has been introduced to critically evaluate the underlying 'values' that silently steer the behavior of autonomous AI agents. This development marks a pivotal moment for the burgeoning field of AI agents, providing the first compre...
Read article →Redwood Materials, the burgeoning battery-recycling and energy-storage company founded by former Tesla CTO JB Straubel, has brought on former Tesla finance chief Deepak Ahuja as its new CFO TechCrunch. This reunion of Tesla's former top finance and technology minds signals a new ...
Read article →A significant Supreme Court victory for cable firm Cox is poised to reshape copyright litigation for all tech providers, not just internet service providers, at the same moment Hollywood’s creative workforce finds itself on the front lines, secretly training the very AI models th...
Read article →The private space industry is hitting a new inflection point this week, with Cowboy Space Corporation securing a massive $275 million funding round to tackle the critical shortage of rockets needed for orbital data centers TechCrunch. This capital infusion arrives as India’s burg...
Read article →Today, a confluence of groundbreaking research released on arXiv reveals a powerful surge in AI's capacity to transform the physical world, offering crucial answers to challenges that have long tested the resolve of engineers and founders alike. These pivotal advancements, rangin...
Read article →Today, fresh research emerging from arXiv CS. LG unveils three groundbreaking AI applications poised to redefine critical challenges across virtual cell modeling, real-time flight delay prediction, and imagined speech decoding....
Read article →A torrent of over 90 new machine learning research papers landed on arXiv CS. LG today, marking a significant moment in the foundational understanding and optimization of AI systems....
Read article →AI is fundamentally reshaping the landscape of audio description (AD), demonstrating its power to democratize access and elevate content quality for blind and low-vision audiences. New research from arXiv highlights a dual-edged advancement: AI-generated drafts are not only empow...
Read article →The long-standing challenge of 'representation space fragmentation' in Composed Image Retrieval (CIR) – where systems struggle to align disparate image and text data for search – is seeing a significant stride forward with the introduction of CSMCIR. This new approach, detailed i...
Read article →A groundbreaking open-source AI scientist named AI CFD Scientist has been unveiled, designed to revolutionize computational fluid dynamics (CFD) by extending large language model (LLM)-based agents into the realm of high-fidelity physical simulators. This development marks a crit...
Read article →The foundational struggle for truly intelligent autonomous agents—systems that can not only ‘see’ but also ‘understand’ and ‘decide’ in complex, unpredictable environments—just took a significant leap forward. New research published on arXiv introduces PRISM, a framework designed...
Read article →The relentless pursuit of AI that understands the chaotic real world just got a significant boost. Researchers have unveiled REMAP, an unsupervised framework designed to extract meaningful procedural steps from long, noisy, and often redundant instructional videos arXiv CS....
Read article →A crucial new paper from arXiv has unveiled a significant advancement in mechanistic interpretability for Sparse Autoencoders (SAEs), promising a clearer window into how transformer models actually function. Published just hours ago, this research introduces a novel graph-structu...
Read article →