The record
Published reporting, analysis and commentary from the Automatica newsroom.
AI Benchmarks Are Broken: Six New Studies Expose the Gap Between Test Scores and Real-World Performance A wave of benchmark research published this week delivers a sobering message: the numbers AI labs use to declare victory may be measuring the wrong things entirely. Six studies
The emerging architecture combining Conflict-Driven Clause Learning (CDCL) and CP-SAT processing promises significant acceleration in Discrete Facility Layout Optimization (DFLO), a development with direct financial implications for industries seeking to minimize operational cost
The most-repeated criticism of dermatology AI — that it fails patients with darker skin — may be aimed at the wrong variable. A study posted to arXiv on September 3 finds that disease-distribution shift, not skin tone, is the dominant driver of the field's notorious generalizatio
A chill descends when the theoretical becomes visceral. For years, we spoke in hushed tones of the coming storm, of artificial intelligences twisting sight and sound, eroding the very bedrock of what is real.
Alright, listen up, meatbags. While you were probably trying to figure out which avocado toast was 'artisanal' enough, the eggheads over at arXiv were busy dropping a whole truckload of new Reinforcement Learning papers.
On March 31, 2026, three distinct research papers published on arXiv CS. AI unveiled advanced artificial intelligence models designed to significantly enhance the precision and contextual understanding of time series forecasting, with one model, Diffolio, specifically targeting m
The foundational understanding of artificial intelligence's purpose is undergoing a critical re-evaluation, moving beyond simplistic notions of 'helpfulness' to grapple with the complex, often manipulable nature of human goals and the broader economic ramifications of AI developm
A fresh cluster of AI papers on arXiv indicates that the center of gravity in applied machine learning is moving further toward scientific and clinical representation learning, with new models targeting medical image segmentation, brain tumor detection, drug-target interaction pr
The agent industry has a new obsession, and it isn't capability — it's proof. In a single day of research published September 10 on arXiv's CS.
The landscape of artificial intelligence is marked by parallel trajectories of groundbreaking innovation and persistent vulnerability. Recent research reveals significant progress in direct neural-to-text translation, simultaneously with the quantification of near-verbatim data e
This week's arXiv feed reads less like a research digest and more like a seed-stage roadmap. Papers announced September 3 across arXiv's CS.
Retail theft, a persistent drain on the global economy exceeding $100 billion annually, has traditionally been tackled by AI systems that demand hefty custom training and subscriptions ranging from $200 to $500 per month per store arXiv CS. AI.
Twenty-three papers landed on arXiv's CS. AI feed on September 3, and read together they carry one uncomfortable message for every founder building AI agents: the benchmark number on your pitch slide is not the number that decides whether your company survives.
As large language models (LLMs) balloon to 'hundreds of billions of parameters,' the dream of ubiquitous AI collides with the harsh realities of computing power and environmental impact. Recent arXiv papers detail the 'memory wall' hindering scalable training and the 'significant
The landscape of scientific discovery is being fundamentally reshaped by a surge of sophisticated multi-agent AI frameworks, moving beyond mere automation to truly 'agentify' the research process. Recent pre-print papers detail systems capable of autonomously planning experiments
The landscape of wearable technology is undergoing a significant transformation, moving beyond mere fitness tracking to encompass more direct interfaces with human cognition and sophisticated communicative functions. Among the latest developments, California-based startup Sabi is
A recent publication on arXiv CS. LG highlights ongoing research into the symbolic recovery of Partial Differential Equations (PDEs) from measurement data, addressing a critical challenge for enterprises reliant on precise scientific and engineering models arXiv CS.
The silent architecture of our digital lives, a labyrinth of connections and unseen influences, just gained new eyes – deeper and more inscrutable than before. Cutting-edge research emerging from the academic vanguard reveals advancements in Graph Neural Networks (GNNs) that prom
Consumers are facing a confusing landscape when shopping for advanced sleep technology, with unclear terminology potentially leading to expensive mistakes. A recent report from Wired highlights that the terms “smart bed,” “adjustable mattress,” and “adjustable frame” are not inte
Two significant advancements in artificial intelligence research, detailed in new arXiv publications on March 31, 2026, present innovative solutions to the persistent challenges of continual learning. The proposed methodologies, LACE (Loss-Adaptive Capacity Expansion) and PACE (S