1841 articles
A new framework explains artificial intelligence's real-world limitations by shifting focus from raw computational power to context management and entropy reduction in human-machine interactions.
Elon Musk is pursuing $79-134 billion in damages from OpenAI and Microsoft over claims the AI company abandoned its nonprofit roots. The high-stakes case heads to jury trial in April after a judge rejected dismissal attempts.
US data centers are on track to consume almost 10% of the country's total electricity by 2030, showing a dramatically sharper increase compared to other major regions worldwide.
Court filings reveal 2017 internal conflicts at OpenAI over leadership control and funding strategy, including Greg Brockman's concerns about Elon Musk's influence and questions about pursuing a billion-dollar valuation while maintaining nonprofit status.
DeepMind CEO Demis Hassabis says today's AI systems still can't match human reasoning. The technology lacks crucial capabilities in long-term planning and understanding how the physical world actually works.
AI data centers worldwide now consume around 30 gigawatts of power—roughly equal to New York State's peak electricity usage on the hottest summer days, according to new research from Epoch AI.
Black Forest Labs' FLUX.2 Klein 9B has claimed the top position among open-weights image editing models on the Artificial Analysis Image Editing Arena, outperforming Alibaba's Qwen Image Edit and other major competitors.
Chinese AI startup DeepSeek pivoted to Nvidia hardware for training its upcoming flagship model after domestic Chinese chips failed to meet performance requirements. The company expects to launch the new model within weeks.
Grok Code Fast 1 matched Claude Opus 4.5's bug detection accuracy in real-world testing, both achieving a 44% detection rate. The free-tier model outperformed Gemini 3 Pro while completing code review tasks in just two minutes.
Lightricks' LTX-2 has overtaken Alibaba's Wan 2.2 A14B to claim the number one spot among open-weights video generation models, according to new Artificial Analysis benchmarks testing both text-to-video and image-to-video capabilities.
The explosive growth of AI-driven data centers is triggering unprecedented electricity demand, pushing up prices for critical commodities including copper, steel, and energy resources.
Google has unveiled TranslateGemma, a suite of open-source translation models powered by Gemma 3 architecture, delivering advanced performance across various devices while supporting over 500 language pairs for seamless multilingual communication.
OpenBMB has released VoxCPM, a tokenizer-free text-to-speech model that clones voices in real-time with streaming inference and LoRA fine-tuning capabilities.
Anthropic's new Economic Index Report reveals how Claude AI performs across different education levels and regions, showing strong productivity boosts on difficult work but declining success rates as tasks get harder.
Google Gemini dominated year-over-year traffic growth among major generative AI platforms in December 2025, according to Similarweb data. Claude, Perplexity, and ChatGPT also showed impressive user gains.
Kawasaki Heavy Industries is rolling out its Nyokkey wheeled humanoid robot across hospitals and care facilities in Japan, handling patrols, guidance, and meal delivery through autonomous navigation powered by hydrogen fuel cells.
The AA-Omniscience benchmark reveals no single AI model dominates software engineering tasks, with Claude Opus 4.5 leading Python and Go while Gemini 3 Pro Preview takes JavaScript and Swift.
Anthropic's Claude Opus 4.5 just claimed the top spot on daily token rankings for the first time, crunching through 149 billion tokens with a massive 59% jump from previous performance.
A new framework called MAXS enables AI agents to look ahead before acting, simulating multiple future action paths and evaluating trajectory stability to avoid short-sighted decisions in complex multi-tool tasks.
A new AI model called UniPercept teaches machines to understand images like humans do, evaluating aesthetics, quality, structure, and texture instead of just recognizing objects.
We use cookies to improve your experience on our site and to show you relevant advertising. To find our more, read our privacy policy and cookie policy
Saad Ullah
Eseandre Mordi
Sergey Diakov
Alex Dudov
Usman Salis
Peter Smith
Marina Lyubimova
Artem Voloskovets
