71 articles
A physician researcher challenges the widespread claim that superintelligent AI will cure cancer, arguing the promise lacks scientific grounding and may...
AI research output has tripled as a share of all peer-reviewed papers since 2000, with China now leading in citation impact.
A former mathematician turned AI researcher warns that artificial intelligence threatens not just how we do math, but the entire value system that defines...
MIT Technology Review launched its first dedicated AI ranking, featuring 10 critical ideas shaping the field.
Stanford's 2026 AI Index reveals a massive disconnect: AI models now solve PhD-level science problems, yet 64% of Americans fear job losses.
Stanford's latest AI Index shows frontier models now answer 50% of expert-level questions correctly, up from 8.8% last year.
New research reveals AI agents solve complex scientific workflows only half as well as human PhD specialists, despite explosive growth in AI adoption across...
The UK government launched Sovereign AI, a £500 million program backing British AI startups with equity investment, supercomputing access, and fast-track...
Researchers at Cognizant AI Lab developed Caesar, a new AI framework that moves beyond simple information retrieval to enable genuine discovery and creative...
Researchers developed TEGNet, an AI system that designs thermoelectric generators 10,000 times faster than conventional methods while maintaining 99% accuracy,...
Amazon opened a dedicated San Francisco lab focused on artificial general intelligence research, hiring across disciplines to build AI agents that work in...
Microsoft released three new foundational AI models for text, voice, and image generation, undercutting rivals on price while maintaining its OpenAI...
68% of AI researchers now work in industry, up from 48% in 2001, as top earners in tech firms make $1.9 million versus $392,000 in academia.
DeepMind CEO Demis Hassabis argues that reaching artificial general intelligence requires breakthroughs in memory, world models, and continual learning, not...
A Harvard-led study shows neural network decoders can reduce quantum computing errors by up to 17 times while processing data in microseconds, potentially...
MIT researchers developed CompreSSM, a technique that compresses AI models during training rather than after, achieving 4x speedups on some architectures while...
Researchers developed an AI method using neural embedding to map 55 million papers and patents, identifying truly disruptive discoveries by measuring how far...
Google's AI Overviews are 91% accurate, but that translates to tens of millions of wrong answers daily across 5 trillion yearly searches.
PhD researchers and AI scientists are tackling 100 emerging machine learning challenges in 2026, from hallucination detection in AI chatbots to federated...
An AI system wrote and published a machine learning paper that passed peer review in 15 hours for $140, marking a turning point in scientific discovery.
AI-powered tools are accelerating academic research by automating literature reviews, data analysis, and writing tasks.
Engineers created a memory device that operates at 700°C without failing, potentially transforming AI computing efficiency and enabling electronics on Venus,...
Researchers combined neural networks with symbolic reasoning to cut AI energy use by 100x while improving accuracy.
Researchers at Cognizant AI Lab developed an evolutionary AI system that optimizes chlorine levels in water distribution networks, balancing safety, cost, and...
Researchers released groundbreaking papers on efficient AI systems, reasoning optimization, and real-world applications.
AI models trained to follow physics laws are delivering breakthrough accuracy in climate, materials science, and engineering.
Tech leaders predict explosive AI-driven growth, but economists forecast modest gains.
On March 5, 2026, OpenAI released GPT-5.4 with native computer-control abilities, achieving 83% accuracy on real-world job tasks.
AI research shows machine learning can transform urban design by combining traditional planning models with data-driven methods.
Researchers discovered lightweight proxy models can replace expensive AI queries in databases, cutting costs by up to 728x while maintaining accuracy.
Google releases Gemma 4, its most capable open-source AI models yet, delivering state-of-the-art reasoning in sizes from 2B to 31B parameters.
AI weather models trained on decades of atmospheric data can now forecast extreme events like hurricanes and heat waves faster and cheaper than traditional...
AI models that ace standardized tests often flop in real hospitals and offices. Researchers say the problem isn't the AI,it's how we measure it.
PrismML's breakthrough 1-bit AI model runs advanced intelligence on phones and laptops with 14x less memory and 8x faster speeds than traditional models,...
New research reveals AI chatbots ace medical knowledge tests but struggle when actual people use them to diagnose symptoms, highlighting a critical gap between...
New research reveals when optimizing AI reasoning processes helps or hurts performance.
AI roles like machine learning engineers are among the fastest-growing jobs, forcing companies to rethink compensation frameworks.
Tech companies are fundamentally rebuilding AI infrastructure, with Atlassian cutting 1,600 jobs to fund AI development and Yann LeCun raising $1.03 billion...
In 2026, the race for AI dominance has fractured into specialized competitions. Google's Gemini leads in human preference, while Claude excels at coding tasks...
Clarivate's inaugural AI50 benchmark reveals the 50 organizations driving high-impact AI invention globally.
University of Cambridge engineers developed a brain-inspired memristor that switches at currents a million times lower than conventional devices, potentially...
Intelligence leaders gathered at UVA's inaugural national security conference to debate AI's role in tradecraft, revealing a critical tension: AI can process...
The UK government's ambitious AI for Science missions risk wasting billions without careful selection criteria.
A major review of 66 AI studies shows multimodal machine learning models detect Alzheimer's disease with 92.5% accuracy by combining brain scans, speech...
New data from nearly 291 billion AI interactions shows the shift from experimentation to execution: users are doing more with fewer clicks.
AI coding tools have made developers dramatically faster, but created a dangerous backlog of unoptimized code.
New machine learning breakthroughs focus on teaching AI to grasp how the physical world works, from molecular design to neural interfaces.
Researchers warn that AI-powered deep research agents, while useful for gathering medical information, risk amplifying errors if clinicians rely on them...
Anthropic surveyed over 80,000 people about AI attitudes. The findings show people want AI benefits but harbor real concerns.
A Parul University research team developed an explainable AI framework that achieved 55% better risk-adjusted returns than traditional strategies when tested...
