AI News

45% of Saudi Internet users now use AI

The Saudi Internet Report 2025 by the Communications, Space and Technology Commission reveals that 45.2% of internet users in Saudi Arabia now utilize AI tools. Adoption is notably higher among women at 52.8% compared to 39.4% for men, and it declines with age, with 58.4% usage among those aged 10-19. ChatGPT ranks as the most downloaded AI application in Saudi Arabia, followed by Google Gemini and DeepSeek, with the local HUMAIN AI app in the tenth position. The report indicates that information searching is the primary use case for AI tools at 80.8%, followed by idea generation (40.1%) and education (37.7%).

💡

Alibaba's Qwen takes on Kimi K3 with open-weight Qwen 3.8, says model is "second only to Fable 5"

How do you foresee Qwen 3.8 impacting the AI market?

Chinese Tech Companies Race to Launch AI Agent-Enabled Smartphones

Chinese Tech Companies Race to Launch AI Agent-Enabled Smartphones

Chinese tech companies are competing to launch new smartphones equipped with AI agents, such as Nubia's Navix Ultra. These phones were showcased at the Global AI Conference in Shanghai, where AI agents can perform tasks like ordering food and booking services through voice commands. These smartphones face market challenges, as platforms like Alibaba and Tencent have restricted AI agents' access to their applications, limiting their capabilities. IDC notes that these obstacles hinder the widespread adoption of new smartphones, even though AI agents represent the future of technology.

Google Deepmind argues video generators already contain the world models computer vision has been missing

Google Deepmind argues video generators already contain the world models computer vision has been missing

Google DeepMind has launched GenCeption, a model that repurposes a video generator for classic vision tasks like depth estimation and segmentation. This model is trained almost entirely on synthetic videos, enabling it to achieve results that match state-of-the-art systems while using significantly less training data. The results from GenCeption contribute to the ongoing debate about whether video generators already contain a kind of universal world model that computer vision has been lacking. This model represents a significant step towards enhancing the efficiency of vision models, opening new avenues for the use of synthetic data.

Moonshot's Kimi K3 outperforms Fable 5 in frontend code but lags far behind in complex math

Moonshot's Kimi K3 outperforms Fable 5 in frontend code but lags far behind in complex math

Moonshot's Kimi K3 has achieved a significant milestone by topping the Code Arena frontend rankings, outperforming both Claude Fable 5 and GPT-5.6 Sol by a considerable margin. Despite its success in coding, Kimi K3 struggles with advanced mathematics, scoring only about 39% on the FrontierMath Tier 4, while models from OpenAI and Anthropic approach nearly 90%. This performance gap highlights the challenges faced by Chinese models in specific areas, necessitating ongoing improvements to compete in the global market.

Huawei Cloud Unveils Enterprise AI at WAIC: Zhou Yuefeng Breaks Down the Steps for Scaling AI Agents Into Production Systems

Huawei Cloud Unveils Enterprise AI at WAIC: Zhou Yuefeng Breaks Down the Steps for Scaling AI Agents Into Production Systems

Zhou Yuefeng, CEO of Huawei Cloud, outlined the company's AI strategy at WAIC 2026. He emphasized that scaling AI agents from demos to production systems requires three coordinated advances: open protocols, reusable industry assets, and a platform approach that enhances agent capabilities. The centerpiece is AgentArts, a full lifecycle platform covering AgentStudio for agile construction, AgentRun for reliable operation, and AgentOps for continuous operations, with openJiuwen as the open-source core framework. Zhou highlighted the stability gap between demonstration and production as a key challenge Huawei is addressing.

AI chatbots reading X-rays can be dangerously confident even when they're wrong

AI chatbots reading X-rays can be dangerously confident even when they're wrong

The RadLE 2.0 benchmark assesses AI models in radiology on their ability to determine when to defer a diagnosis to human experts. Many models exhibit high confidence in incorrect findings, underscoring the gap between AI and human radiologists. Studies indicate that human radiologists still outperform AI models in diagnostic accuracy. It is crucial for AI systems to learn when to refrain from making a diagnosis, highlighting the need for enhanced learning mechanisms.

Business

UAE firms locked into AI vendors, IBM finds

Business