Welcome to the AI Daily section! Here is your daily guide to exploring the world of artificial intelligence, where we present the hottest content in the AI field every day, focusing on developers to help you understand technical trends and innovative AI product applications.
Discover fresh AI products here: https://top.aibase.com/
1. Challenging Google! OpenAI Launches SearchGPT, Inviting 10,000 Testers for Initial Testing
OpenAI has introduced SearchGPT, an AI-powered search engine that organizes and summarizes search results, unlike traditional search engines. Currently in the prototype stage and supported by GPT-4, it is open to 10,000 test users. OpenAI collaborates with third parties to build search results and plans to integrate search functionality into ChatGPT.
【AiBase Summary:】
🔍 SearchGPT is an AI-powered search engine that organizes and summarizes search results.
🚀 Currently in the prototype stage, supported by GPT-4, open to 10,000 test users.
💡 OpenAI plans to directly integrate search functionality into ChatGPT, competing with Google and emphasizing content collaboration and clear information attribution.
Details link: https://chatgpt.com/search
2. Zhipu AI Launches AI-Generated Video Product Qingying
Zhipu AI has launched the Qingying AI large model, capable of generating high-precision videos from any text. Users can input text and select a style to generate high-quality videos. Qingying is now available on the Qingyan App, supporting text-to-video and image-to-video, and also offers a "Photos Come Alive" mini-program. CEO Zhang Peng mentioned that the video generation model CogVideoX at the core of Qingying integrates text, time, and space dimensions, improving inference speed. Users can experience Qingying on the Zhipu Qingyan PC/APP, turning inspiration into artistic video creations.
【AiBase Summary:】
🎥 Qingying is an AI large model launched by Zhipu AI, capable of generating high-precision videos, supporting text-to-video and image-to-video.
💡 Based on the new DiT model architecture, Qingying integrates text and video content, improving instruction following ability and content coherence.
🚀 CogVideoX is the video generation model at the core of Qingying, integrating text, time, and space dimensions, improving inference speed, with future plans for higher resolution and longer duration video generation capabilities.
Details link: https://top.aibase.com/tool/qingying-ai-shipinshengchengfuwu
3. ByteDance Releases Doubao Image-to-Image Model, Doubao Large Model Exceeds 500 Billion Tokens Daily Usage
At the 2024 AI Innovation Tour event in Chengdu, Volcano Engine announced that the Doubao large model exceeds 500 billion tokens daily usage, with a 22-fold increase in daily usage by customers. Vice President Zhang Xin stated that Volcano Engine is developing towards intelligence, industry, and regionalization to help businesses achieve innovation. The latest capabilities of the Doubao large model include upgrades in visual images, voice synthesis, and voice cloning.
【AiBase Summary:】
🚀 Doubao large model exceeds 500 billion tokens daily usage, with a 22-fold increase in daily usage by customers.
🔍 The Doubao Image-to-Image model and Doubao Text-to-Image model excel in retaining original image features and enhancing image quality.
🔊 The Doubao Voice Synthesis model and Doubao Voice Cloning model have improved in expressing emotions and reproducing speaker voice characteristics.
4. AI Video Generator Runway Exposed for Using Pirated YouTube Content for Training
This article exposes the scandal of Runway's Gen-3 Alpha video generator using pirated content, sparking copyright controversies. AI companies frequently infringe on copyright laws, and legislators are re-examining copyright regulations to adapt to new technological challenges.
【AiBase Summary:】
📊 Runway video generator exposed for using pirated content—sparks copyright controversy
🛡️ AI companies frequently infringe on copyright laws—copyright disputes become a bottleneck for AI development
📜 Legislators re-examine copyright regulations—laws and copyright usage policies constantly updated
5. No More Headline Refugees! Bilibili Launches AIGC Recommended Ad Headline Feature
In this era where creativity reigns, Bilibili's AIGC recommended ad headline feature injects new vitality into ad creation. With AI-generated 10 top-notch headlines, the creative process becomes simpler and more efficient, adding the possibility of enhancing ad effectiveness.
【AiBase Summary:】
🔑 Creativity is king, and headlines are key. AIGC recommended ad headline feature makes creation simpler and more efficient.
🤖 Behind the AI master is extensive data training. Generates diverse, eye-catching headlines.
🚀 Continuously optimize the AI master for more precise, targeted headlines. Boldly expands the freedom of ad creation.
6. Instant AI Search Wonder Wenwen Xiaoyuzhou is Here
Instant App has launched an AI search feature based on Xiaoyuzhou—Wenwen Xiaoyuzhou, focusing on deep mining of audio content, providing in-depth discussions and unique insights. The retro color scheme and personalized recommendation function are its features, making search results richer and more diverse, closer to user needs.
【AiBase Summary:】
🔍 Wenwen Xiaoyuzhou is an AI search feature based on Xiaoyuzhou, focusing on audio content mining.
🎧 Provides in-depth discussions and unique insights, recommending relevant audio content.
🎨 Retro color scheme, personalized recommendation function, rich and diverse search results, close to user needs.
Details link: https://top.aibase.com/tool/wenwenxiaoyuzhou
7. The "AI Agent" of Translation! ByteDance Launches the End-to-End Speech Synchronization Translation System CLASI
CLASI, the end-to-end speech synchronization translation system launched by ByteDance, brings innovation to global communication. It combines language models and information retrieval systems to achieve accurate and fast translation, with a contextual memory function that surpasses human translators. Although not perfect, CLASI provides efficient translation services through clever coping abilities. CLASI's emergence opens up a new world for cross-language communication, bringing a gentle revolution to human communication methods.
【AiBase Summary:】
🌐 CLASI is an end-to-end speech synchronization translation system, combining language models and information retrieval systems for accurate and fast translation.
🧠 CLASI has a contextual memory function, connecting previous content to ensure translation coherence, surpassing human translators.
🔍 CLASI uses clever coping abilities to guess meaning and provide reasonable translations, outperforming commercial and open-source systems in conveying effective information.
Details link: https://top.aibase.com/tool/clasi
8. Wuhan University, in Collaboration with China Mobile Jiutian Artificial Intelligence Team, Open-Sources the Audio-Visual Speaker Recognition Dataset VoxBlink2
Wuhan University, in collaboration with China Mobile Jiutian Artificial Intelligence Team and Kunshan Duke University, has open-sourced the VoxBlink2 audio-visual speaker recognition dataset based on YouTube data, exceeding 110,000 hours, the largest publicly available audio-visual speaker recognition dataset. This dataset enriches the open-source speech corpus, supporting the training of voiceprint large models.
【AiBase Summary:】
🔍 The dataset size exceeds 110,000 hours, including 9,904,382 high-quality audio and video clips from 111,284 YouTube users.
🔬 The dataset underwent multi-step data mining, including candidate preparation, face extraction & detection, face recognition, active speaker detection, etc., with accuracy improved to 92%.
🛠 VoxBlink2 open-sources voiceprint models of various sizes, including 2D convolution models based on ResNet and temporal models based on ECAPA-TDNN, as well as the super-large model ResNet293, performing excellently on the Vox1-O dataset.
Details link: https://VoxBlink2.github.io
9. Google Gemini Major Update: Multilingual Support, Performance Improvement, Open to Teenagers
Google announces a comprehensive upgrade of its AI chatbot Gemini, including multilingual support, performance improvement, and opening to teenagers. This update will enhance user experience, reduce operating costs, increase transparency, expand application scenarios, demonstrating Google's ambition and determination in the AI field.
【AiBase Summary:】
🌐 Multilingual support: Gemini 1.5 Flash supports 40 languages, covering 230 countries and regions, improving quality and response speed.
🔍 Context window expansion: Gemini's context window expands to 32,000 tokens, supporting longer text processing and file upload functions.
🚀 More extensive application scenarios: Gemini's functions will expand to Messages app integration, mobile app promotion, and opening to teenagers.
10. Easy Tuning! Microsoft Introduces Serverless Fine-Tuning Feature for Phi-3 Small Language Model
Microsoft introduces a serverless fine-tuning feature for the Phi-3 small language model, providing developers with an easy way to adjust and optimize model performance. This move will further promote the development and popularization of AI applications.
【AiBase Summary:】
📈 Serverless fine-tuning feature: Developers can easily adjust the Phi-3 model without managing servers, improving performance.
💰 Cost-effective Phi-3 model: Provides high performance at low cost, suitable for various enterprise application scenarios.
🤖 Intense market competition: Microsoft competes with AI providers like OpenAI, driving the development of the AI industry.
Details link: https://azure.microsoft.com/en-us/blog/announcing-phi-3-fine-tuning-new-generative-ai-models-and-other-azure-ai-updates-to-empower-organizations-to-customize-and-scale-ai-applications/
11. Musk Seeks Tesla Board Approval to Invest $5 Billion in xAI
Musk plans to invest $5 billion in AI startup xAI, potentially sparking conflicts of interest among tech companies. Tesla is transforming into a robotics and AI company, with Musk promising an autonomous robotaxi and humanoid robot fleet. xAI was founded in July last year, valued at $18 billion, with Musk catching up to competitors like OpenAI and Anthropic.
【AiBase Summary:】
🚀 Musk plans to invest $5 billion in xAI, potentially sparking conflicts of interest among tech companies.
🤖 Tesla is transforming into a robotics and AI company, with Musk promising an autonomous robotaxi and humanoid robot fleet.
💰 xAI was founded in July last year, valued at $18 billion, with Musk catching up to competitors like OpenAI and Anthropic.
12. Google AI's Geometric Super Evolution: Crushing Human Players in IMO with a 19-Second Solving Speed
