Google has launched Vids in Workspace Labs, allowing users to create presentation videos by inputting documents, slides, voiceovers, and recorded video content into a timeline. Vids is not to be confused with AI tools like Sora, as it generates presentation slides based on content descriptions. Gemini automatically inserts stock materials, generates scripts, and provides AI voiceovers.

Groundbreaking SECBOOST technology by Google research team revolutionizes boosting methods in zeroth-order optimization, showing superior performance and promising new directions in machine learning.

InternLM has released InternLM2.5-7B-Chat, a large-scale language model in GGUF format with enhanced flexibility and efficiency for various applications. With 7 billion parameters, optimized for chat scenarios, it outperforms industry competitors. It excels in long-context tasks, achieving high scores in benchmark tests. Equipped with LMdeploy framework, it offers high efficiency in model deployment.

Meta's ChatGPT AI chatbot receives mixed reviews, but generative AI is considered key for future interactions. Company X promotes its Grok chatbot through various channels to enhance user experience, with plans to add analytics and analysis options. Elon Musk's influence drives Company X to expand AI tools, despite limitations to premium subscribers.

Anthropic's AI model's training cost could hit $1 billion, escalating to $10 billion or $100 billion by 2027. AI's exponential growth requires a tenfold increase in hardware performance annually, with AI chip demand soaring. Energy supply and infrastructure must keep pace to meet escalating power demands.

In current energy tech, training advanced AI models is unsustainable. Google DeepMind's JEST revolutionizes model training, boosting speed by 13x and energy efficiency by 10x. JEST optimizes batches, reducing costs and environmental impact. AI training efficiency can be improved further by exploring JEST's potential.

OpenAI restricts Chinese users' access to AI models via API, exempting Microsoft Azure clients in China. The move aligns with policies to limit Chinese access due to geopolitical concerns. China leads in AI patents, alarming the US.

Apple Intelligence will debut on iPhone, Mac, and iPad only. Siri won't integrate with HomePod yet. AI desktop robot marks Apple's smart home innovation. The robot will disrupt traditional technology with advanced AI algorithms for interactive experiences. HomePod may still have AI upgrade plans. Market excited for Apple's new desktop robot, expecting it to drive smart home industry towards a more intelligent future.

Samsung plans to expand Live Translate feature to WhatsApp & other communication apps, enabling real-time translation during messaging & calls for seamless cross-language communication. Official confirmation pending.

Hebbia, an AI startup, raised $130M in Series B funding. Its flagship product Matrix stands out for handling complex queries in datasets. CEO Sivulka aims to make AI work like humans, attracting clients like top asset management firms and the US Air Force. Hebbia's growth is evident, with revenue growing 15 times in 18 months.

Meta AI researchers released MobileLLM, a compact language model optimization approach for smartphones. It prioritizes model depth, implements embedding sharing, and achieves 2.7-4.3% better performance. The 350 million parameter MobileLLM shows comparable accuracy to larger models, highlighting efficiency. Despite not being publicly available, Meta has open-sourced the pre-training code for further research and potential advancements in AI capabilities on personal devices.

Tsinghua University released CodeGeeX4-ALL-9B model with 9.4 billion parameters, significantly improving code generation performance. It excels in multi-language code generation and interaction efficiency, leading the industry in practical applications. With user-friendly design and flexible deployment on GPUs/CUPs, it simplifies software development, enhances coding efficiency, and promotes innovation in the global software industry.

WhatsApp is developing a new generative AI feature for personalized avatars in various scenarios. Users can create avatars with user images and text prompts using the Llama AI model by Meta. The feature is optional and requires manual activation with the ability to delete reference images. The release date is currently uncertain.

A recent study reveals that AI like ChatGPT can create jokes as funny as humans. Participants preferred ChatGPT jokes over human jokes, showing AI's potential in humor creation. This raises concerns about AI's impact on human creativity in the entertainment industry.

Grok fueling discussion on X platform's cryptocurrency payments, hinting at XRP's role. Speculation on Elon Musk testing XRP payments. XRP and XLM as payment competitors with uncertain plans. Musk's affinity for Dogecoin noted. X holds payment licenses in 31 US states. Community eagerly awaits X's payment strategy reveal.

At WAIC 2024, Bilibili revealed its latest AI advancements in content creation, including digital avatars, dynamic comics, and AI voice libraries like Luo Tianyi. The platform's AI content attracts 80M+ monthly users, especially younger demographics, bridging the gap between the AI industry and users. Bilibili collaborates with content creators and brands to promote AI technology through creative marketing and comprehensive support tools for content creation, enhancing efficiency and diversity.

The latest Android WhatsApp beta version introduces a feature integrating Meta AI, allowing users to send photos to AI for questions or editing. This enriches communication and showcases Meta's exploration in image technology.

Siri to see major upgrades in spring 2025 with Apple Intelligence AI opening to all users through iOS 18.4. Features may include auto-ID retrieval but AI-driven emoji or writing assistance may not be ready yet. New Siri interface and ChatGPT integration expected in fall, with rumored AirPods with built-in cameras on the horizon.

Kyutai's Moshi sets a new benchmark in voice AI, displaying over 70 emotions and speaking styles. Its realistic performance in real-time conversations surpasses traditional systems, enhancing user experience.

Samsung will unveil its latest foldable phone at the upcoming Galaxy Unpacked event. Rumors suggest the Galaxy Ring may launch alongside new Galaxy Watch and Galaxy Buds models. Improved AI capabilities are also expected to be showcased.