Welcome to the [AI Daily] column! Here's your guide to exploring the world of AI every day, and every day we present you with hot content in the field of AI, focusing on developers, helping you gain insight into technology trends and understand innovative AI product applications.
Fresh AI productsClick to learn: https://top.aibase.com/
1. Kimi launched a multi-function voice call mode that supports changing the voice and adjusting the speaking speed
Kimi's AI assistant recently launched the highly anticipated voice call feature, bringing a rich interactive experience to users. The new feature combines diversified applications and personalized experiences to improve communication efficiency and meet personalized needs. Functional applications such as English sparring and mock interviews have been deepened for specific scenarios, providing innovative solutions for learning and the workplace. Introduce voice cloning technology to add interest, continue to pay attention to user feedback, and optimize and expand the application scope of voice services.
【AiBase Summary:】
🌟 Kimi Assistant has updated its voice call feature to improve user interaction.
🎤 Support for real-time subtitles and adjustable sound for more natural interactions.
🤖 Introduce sound cloning technology to increase interest and application scenarios.
2. Zhipu CogView3-Plus model open source Wensheng graph technology ushered in a new era
The technical team of Zhipu has released CogView3 and its upgraded CogView3-Plus-3B, marking a new stage of AI-assisted art creation. CogView3 delivers incredible performance, 77% faster than SDXL, and inference is only one-tenth the speed of SDXL. CogView3-Plus introduces the DiT framework and the text-image joint attention mechanism to improve performance and reduce costs. The open source repository has been opened, which promotes the rapid development of the field of AI image generation, and the application of Wensheng graph technology has broad prospects.
【AiBase Summary:】
🚀 CogView3及CogView3-Plus-3B开源，标志AI辅助艺术创作新阶段。
💡 CogView3 delivers incredible performance, 77% faster than SDXL, and inference is only one-tenth the speed of SDXL.
💻 CogView3-Plus introduces the DiT framework and the text-image joint attention mechanism to improve performance and reduce costs.
Details: https://top.aibase.com/tool/cogview3Plus
3. Adobe launched the Firefly AI video model, and Photoshop added multiple AI features
Adobe unveiled the Firefly AI video model and several enhanced Photoshop tools at the latest MAX conference, bringing exciting new AI features to users. The Firefly video model allows users to create or modify video content with text prompts, giving creators a whole new creative space. At the same time, Photoshop's new AI tools, such as the Remove tool and the Generate Fill feature, make it easier for users to edit images. Adobe is committed to training models using only licensed content and is rolling out content credentials to increase transparency.
【AiBase Summary:】
📹 The Firefly AI video model allows users to create and modify video content from text, giving creators a whole new creative space.
🖼️ Photoshop has added several new AI tools, such as the Remove tool and Generate Fill, to help users edit images more easily.
🔍 Adobe is committed to training models using only licensed content and is rolling out content credentials to increase transparency.
Link: https://blog.adobe.com/en/publish/2024/09/11/bringing-gen-ai-to-video-adobe-firefly-video-model-coming-soon
4. ChatGPT makeover: the new interface introduces a search function
OpenAI recently made a major update to ChatGPT, transforming it from a mere conversational assistant to a comprehensive messaging platform. UI updates include new Canvas functionality and mobile tooltip locations to make the user experience even more convenient. The introduced SearchGPT feature gives users access to real-time web data and works with reputable publishers to ensure information reliability.
【AiBase Summary:】
🔍 New interface design: Added the Canvas function and moved the tooltip position to improve the user experience.
🌐 Introducing the SearchGPT feature: press / key to activate, access real-time web data, complete with inline references and links.
💡 Easy to use: Intuitive operation, provide a source answer within seconds of asking a question, and allow you to ask questions to explore the topic in depth.
5. Volcano Engine released a large model training video preprocessing solution
At the recent Video Cloud Technology Conference, Volcano Engine released a large model training video preprocessing solution, which has been successfully applied to the bean bag video generation model, marking a major progress in AI video generation technology.
【AiBase Summary:】
🔑 AIGC and multimodal technology have profoundly changed the user experience
🔑 Volcano Engine explores the integration of AI large models and video technology to provide a full range of solutions
🔑 The preprocessing scheme developed by Volcano Engine optimizes the algorithm and engineering level to improve the efficiency of model training
6. Midjourney is about to launch an online editor, and the V7 version and personalization features are on the way
Midjourney recently announced a series of exciting updates and upcoming new features, showcasing the company's rapid development and innovation in the field of AI image generation. The team is actively working on multiple projects to focus on improving quality and developing core features, rather than rushing to release new products.
【AiBase Summary:】
🚀 Midjourney is launching a new online editor that allows users to import images and edit them with depth maps and image inpainting.
💡 The team is refining the user interface to distinguish new features from existing image features, and discussing what features will be included when it is first released.
🔧 Midjourney is developing new tools to make it easier for users to manage and switch between multiple personalization configurations, introducing a more granular style personalization system.
7. Microsoft is building an AI companion that can "see" and "remember" all user behaviors
Recently, Microsoft is working on developing an AI companion that can observe user behavior in real time, aiming to build deep interactions. This reflects the basic human need for confiding, understanding and companionship. At the same time, a new feature in Windows, Recall, digs deep into a user's digital history to recreate past experiences. Zoom introduces AI cloning to allow users to send digital versions of themselves to meetings to increase productivity.
【AiBase Summary:】
🌟 Microsoft is developing an AI companion that can observe user behavior in real-time, with the aim of building deep interactions.
📂 A new feature in Windows, Recall, digs deep into a user's digital history to recreate their past experiences.
💬 Zoom introduces AI cloning to allow users to send digital versions of themselves to meetings to increase productivity.
8. Microsoft's vice president of AI jumped to OpenAI, which will be committed to AGI research
Sebastian Bubeck, Microsoft's vice president of AI, has left Microsoft to join OpenAI to focus on artificial general intelligence (AGI). This shift marks a career change for Bubeck and adds strong support to OpenAI's development.
【AiBase Summary:】
🧠 Bubeck left Microsoft to join OpenAI, focusing on AGI research.
🤝 Microsoft and OpenAI have a complex relationship, both as investors and competitors, and the two sides want to continue working together.
🚀 Bubeck's addition will propel OpenAI's breakthroughs in small language models and AGI research
9. Amazing! The AI model DIAMOND can simulate Counter-Strike, and it can be run in a single RTX3090
The AI model DIAMOND developed by the researchers is capable of simulating the game Counter-Strike: Global Offensive and runs on an Nvidia RTX3090 graphics card at 10 frames per second. Despite some drawbacks, it is expected that model performance will be improved in the future by increasing data and computing power.
【AiBase Summary:】
🎮 DIAMOND模型能模拟CS:GO游戏，运行速度达每秒10帧。
📊 With only 87 hours of game data training, it can still simulate complex game scenarios.
🐛 There are bugs such as infinite jumps and path deviations that cause the simulation to crash, but there is potential for improvement in the future.
Details:https://diamond-wm.github.io/
10. Nvidia Blackwell GPU demand has skyrocketed, and the supply is completely sold out in the next 12 months
NVIDIA's latest-generation Blackwell GPUs have sparked unprecedented demand, with CEO Jensen Huang revealing that supplies are sold out over the next 12 months, reaffirming the market's thirst for high-performance computing.
【AiBase Summary:】
💡Blackwell GPUs are in explosive demand and are completely sold out within the next 12 months.
🔍 Major customers include big technology companies such as AWS, Google, etc., driving the market growth.
📈NVIDIA expects to ship billions of dollars of Blackwell GPUs in 2024, with significant growth in data center revenue.
11. Nvidia's stock price hit another all-time high, with a market capitalization approaching $3.4 trillion
Investors are excited about Nvidia's stock price hitting another all-time high this week, with a market capitalization approaching $3.4 trillion. CEO Jensen Huang's personal fortune has exceeded $121.5 billion, and the demand for the new generation of Blackwell chips is hot. Nvidia firmly dominates the AI chip market, and its market value is second only to Apple, showing strong strength.
【AiBase Summary:】
🚀 Nvidia shares hit a new high of $138.07, up nearly 180% for the year.
📊 The company has a market share of 70% to 95% in the AI chip market, with a market capitalization of $3.4 trillion.
💡 CEO Jensen Huang said demand for the new generation of Blackwell chips is "crazy" and has been sold out until next year.