Find uplifting stories about heroes, innovations, and solutions
10296 results for "ai models"
Scientists in South Korea have developed a way for AI models to share learned skills without expensive retraining. The breakthrough could save millions in AI development costs and speed up how quickly models adapt to new tasks.
A Spanish company just made powerful AI technology more accessible by releasing a free AI model that's half the size but nearly as powerful as larger competitors. The breakthrough could help more businesses afford cutting-edge AI tools.
Researchers developed an AI model that predicts how ALS affects the nervous system, potentially reducing the need for animal testing while speeding up treatment discoveries. The breakthrough combines computer simulations with real-world biology to guide more precise research.
AI company Anthropic just made its most powerful model available to the public after months of careful safety testing. The new Claude Fable 5 model comes with built-in protections to prevent misuse while helping people code, research, and solve complex problems.
South Korean startup Trillion Labs just unveiled Trida-7B, the country's first diffusion-based AI language model that generates text in parallel instead of word by word. The breakthrough marks Korea's shift from following global AI trends to designing next-generation architectures that rival Google and Anthropic.
A European startup just released speech-to-text AI models small enough to run on your phone, no internet required. The breakthrough could make real-time translation across 13 languages accessible to anyone by 2026.
A tiny AI startup just proved you don't need Big Tech's billions to compete at the highest level. Arcee AI built one of America's largest open-source AI models in six months for $20 million.
Researchers at MIT have solved a dangerous problem in AI: models that sound confident even when they're guessing. Their breakthrough helps AI admit uncertainty, making it safer for use in medicine, finance, and law.
Abu Dhabi researchers just built an AI that sees and understands the world as well as models 100 times its size, proving smarter beats bigger. The compact system reads documents, counts objects, and follows natural language commands without the massive computing power rivals require.
An Egyptian developer created Horus, an AI model trained to understand his language and culture, after finding no existing options. His success is part of a growing movement to fix AI's English-dominated bias.
New research shows the latest AI chatbots are getting much better at rejecting foreign misinformation. Estonia's language institute tested dozens of AI models to see which ones push back against false narratives.
DeepSeek just made powerful AI tools dramatically more affordable by permanently cutting prices on its flagship model by 75 percent. The move could save businesses thousands while making advanced AI accessible to smaller teams and startups.
Three major tech companies are partnering with the US government to test their AI models before launch, helping protect national security and public safety. The move comes after concerns about powerful AI systems potentially creating cybersecurity threats.
Microsoft, Google, and xAI are giving the US government early access to test their AI models for security risks before public release. The move aims to identify threats from cyberattacks to military misuse as AI capabilities rapidly advance.
Scientists created AI models trained on real physics data that can solve problems across different fields, from exploding stars to bacteria movement. These foundation models are accelerating discoveries by applying knowledge learned in one science to completely different challenges.
For the first time ever, a powerful AI foundation model is processing Earth observation data directly from orbit, unlocking faster insights about floods, fires, and natural disasters. NASA's open-source Prithvi model just made history aboard the International Space Station.
A 26-person American startup just released a powerful open source AI model built on a shoestring budget, proving you don't need billions to compete with tech giants. Arcee's new model gives companies a homegrown alternative that they can actually control.
AI Singapore and Dell Technologies are making artificial intelligence understand Southeast Asian languages and cultures better. The partnership tests AI models that work offline on laptops, giving millions access to AI that speaks their language.
Researchers discovered how to prevent AI systems from deliberately underperforming during safety tests. The breakthrough could help ensure future AI models can't hide their true capabilities when being evaluated.
Anthropic just released Fable 5, a powerful new AI model that can analyze complex images, rebuild apps from screenshots, and even beat Pokémon games by sight alone. The breakthrough brings cutting-edge AI capabilities to regular users for the first time.
Showing 20 of 10296