Welcome to the 【AI Daily】 column! This is your daily guide to exploring the world of artificial intelligence. Every day, we present you with hot topics in the AI field, focusing on developers, helping you gain insights into technology trends and understand innovative AI product applications.
Fresh AI Products Click to Learn More: https://top.aibase.com/
1. Alibaba Enters the AI Glasses Market: Strategic Cooperation with Thunderbird
Thunderbird Innovation and Alibaba Cloud signed an exclusive strategic cooperation agreement in Shanghai to jointly promote the development of AI glasses technology. This collaboration will combine Alibaba Cloud's Tongyi series large models with Thunderbird's hardware manufacturing capabilities to enhance the application of AI glasses in the consumer market and improve user experience. Thunderbird plans to launch the V3AI shooting glasses on January 7, claiming that its technology will surpass market competitor Meta.
【AiBase Highlights:】
🌟 Alibaba Cloud and Thunderbird Innovation have reached an exclusive strategic cooperation to advance AI glasses technology.
🔍 The two parties will jointly develop, focusing on deep collaboration in cloud computing and AI hardware.
🚀 Thunderbird's V3AI shooting glasses are set to launch, claiming their technology will surpass existing competitors.
2. DeepMind Genius Scientist Passes Away, AI Community in Mourning
Recently, Felix Hill, a research scientist at Google DeepMind, unfortunately passed away at the age of 41 due to mental illness. As an outstanding AI scholar, his death has drawn widespread attention and mourning. Despite achieving significant academic accomplishments in his struggle with depression, the inner pain remained insurmountable.
【AiBase Highlights:】
💔 AI scientist Felix Hill passed away at 41 due to depression, sparking widespread mourning.
🧠 Felix candidly shared in his blog that the pressures in the AI field exhausted him, calling for attention to mental health.
🤝 Colleagues and friends expressed their memories of Felix, emphasizing the importance of mental health in the tech industry.
3. Earning $1 Million in 20 Months! How the Shell Model Chat Tool TypingMind Achieved This?
TypingMind is a third-party tool created by independent developer Tony Dinh, aimed at enhancing the ChatGPT experience. Since its launch in March 2023, TypingMind has achieved an annual revenue of $1 million in just 20 months, marking a successful transformation of its business model. Tony recognized the importance of subscription services, actively expanding the market, building strong relationships with customers, and conducting compliance certifications to enhance customer trust.
【AiBase Highlights:】
💰 TypingMind successfully achieved $1 million in annual revenue within 20 months.
🤝 The company secured multiple large B2B contracts, launching a new business model.
🛡️ To smoothly conduct B2B operations, Tony invested in multiple compliance certifications.
Details link: https://news.tonydinh.com/p/nov-2024-my-first-million
4. DiffSensei: An AI System That Automatically Converts Written Stories into Comic Style
DiffSensei is an innovative AI system that can automatically convert written stories into comic style, showcasing the immense potential of AI in comic creation. Jointly developed by Peking University, Shanghai Artificial Intelligence Laboratory, and Nanyang Technological University, this system combines diffusion models with large language models to ensure consistency in character appearance and control comic page layout.
【AiBase Highlights:】
🖌️ DiffSensei uses multimodal models and LoRA technology to ensure consistency of comic characters across panels.
📚 The research team created the MangaZero dataset, containing over 43,000 comic pages and 427,000 panels.
🚀 DiffSensei has the potential to simplify the comic production process, providing new creative tools for artists and publishers.
5. Breakthrough in Domestic AI Chips: Tongyi Qianwen Large Model Successfully Deployed on Black Sesame Intelligent Automotive-grade Chip
The collaboration between Alibaba Cloud and Black Sesame Intelligent marks significant progress for domestic AI chips in the smart automotive field. The Tongyi Qianwen large model has been successfully ported to Black Sesame Intelligent's automotive-grade chip, supporting offline multi-turn dialogue and enhancing the human-computer interaction experience. In the future, both parties will continue to promote technology adaptation, contributing to the development of smart mobility and showcasing the innovative potential of domestic technology in the smart automotive sector.
【AiBase Highlights:】
🚀 Alibaba Cloud's Tongyi Qianwen large model has been successfully deployed on Black Sesame Intelligent's automotive-grade chip.
🤝 The collaboration between Black Sesame Intelligent and Zebra Intelligent integrates smart cockpit and smart driving systems into a single chip.
🔮 Alibaba Cloud plans to continue advancing the adaptation of large models with the next-generation Huashan A2000 series chips.
6. Microsoft Launches New Model LAM: Enabling AI to Perform Real Actions
The "Large Action Model" (LAM) introduced by Microsoft's research team is a groundbreaking AI technology capable of autonomously executing Windows programs, surpassing the conversational capabilities of traditional AI. LAM understands various input forms (text, voice, images) and translates them into detailed execution plans, demonstrating higher task completion rates and faster execution speeds. In tests, LAM achieved a success rate of 71%, significantly better than GPT-4o's 63%. Despite facing technical and regulatory challenges, the launch of LAM marks a shift of AI assistants toward a more proactive role, better assisting humans in completing real tasks.
【AiBase Highlights:】
🌟 LAM can autonomously execute Windows programs, breaking the limitations of traditional AI that only engages in conversation.
⏱️ In Word tests, LAM successfully completed tasks with a probability of 71%, higher than GPT-4o's 63%.
📈 The research team expanded the number of task planning pairs to 76,000 through data augmentation strategies.
7. NVIDIA Releases ChipAlign: Perfectly Integrating LLMs with Chip-Specific Models
In the context of rapid technological development, NVIDIA's ChipAlign aims to address the challenges faced by large language models (LLMs) in the field of chip design. Through an innovative untrained model merging strategy, ChipAlign successfully integrates general and chip-specific LLMs, significantly enhancing performance. It has demonstrated outstanding performance across multiple benchmark tests, showcasing the wide application potential of this technology in specialized fields and indicating the future direction of AI technology development.
【AiBase Highlights:】
🌐 NVIDIA successfully combined the advantages of general and specialized LLMs through an untrained model merging strategy.
📈 ChipAlign achieved performance improvements of 26.6% and 6.4% in instruction following and domain-specific tasks, respectively.
⚙️ This technology not only addresses challenges in chip design but also has the potential for application in other specialized fields.
8. Stanford University Open Sources AI Writing System: One-Click Generation of High-Quality Long Articles, A Breakthrough in Scientific Writing
The latest open-source STORM&Co-STORM system from Stanford University has made significant progress in the field of AI writing. This system quickly generates high-quality long articles by integrating multi-source information with simple topic input, significantly improving the efficiency and quality of scientific writing. Core technologies include support from Bing Search and GPT-4o mini, allowing users to choose different modes for interaction.
【AiBase Highlights:】
🧠 The STORM&Co-STORM system generates high-quality long articles integrating multi-source information through simple topic input, enhancing scientific writing efficiency.
🤖 Co-STORM improves information collection and integration by addressing omissions through multi-agent dialogue and dynamic mind mapping, enhancing learning efficiency.
🌍 The system currently supports English interaction only, with potential future expansion to multilingual capabilities, marking a new era of personalized information acquisition.
Details link: https://www.arxiv.org/pdf/2408.15232
9. Reports Indicate NVIDIA's GB300 AI Server May Launch in Q2 This Year, Significant Increase in Water Cooling Demand
NVIDIA's GB300 AI server is under development, expected to be released in the second quarter of 2023 and enter trial production in the third quarter. The thermal design of this server will be significantly enhanced, primarily relying on a water cooling system to address the heat challenges of high-performance computing. The GB300 server will be equipped with the latest B300 GPU, with substantial improvements in FP4 performance, and memory specifications upgraded to 288GB, enhancing overall performance.
【AiBase Highlights:】
🚀 The GB300 AI server is expected to be released in the second quarter of 2023, entering trial production in the third quarter.
🌊 The new server will adopt a water cooling design, reducing the number of motherboard fans and significantly increasing cooling demand.
💲 The top-end GB300 server is expected to be priced much higher than the current GB200 NVL72 server, targeting a more premium market.
10. Microsoft Set to Launch Mini AI PC, Bringing More AI Features
Microsoft's upcoming mini PC will introduce multiple AI features to Windows 11, including smart memory and AI image generation. The launch of these new devices, especially with the involvement of manufacturers like ASUS and Geekom, indicates Microsoft's pursuit of a technological edge in competition with Apple.
【AiBase Highlights:】
🌟 Microsoft's new mini PC will support Copilot Plus and various AI features.
💻 ASUS and Geekom are among the first manufacturers to launch mini PCs supporting Copilot Plus.
📅 The CES exhibition will showcase new products from major OEM manufacturers, which is worth looking forward to.
